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 or $wgRequest. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.4';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  72         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  73         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  74         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  75                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/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         var $mShowToc, $mForceTocPosition;
 152
 153         /**
 154          * @var User
 155          */
 156         var $mUser; # User object; only used when doing pre-save transform
 157
 158         # Temporary
 159         # These are variables reset at least once per parse regardless of $clearState
 160
 161         /**
 162          * @var ParserOptions
 163          */
 164         var $mOptions;
 165
 166         /**
 167          * @var Title
 168          */
 169         var $mTitle;        # Title context, used for self-link rendering and similar things
 170         var $mOutputType;   # Output type, one of the OT_xxx constants
 171         var $ot;            # Shortcut alias, see setOutputType()
 172         var $mRevisionObject; # The revision object of the specified revision ID
 173         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 174         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 175         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 176         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 177
 178         /**
 179          * @var string
 180          */
 181         var $mUniqPrefix;
 182
 183         /**
 184          * Constructor
 185          *
 186          * @param $conf array
 187          */
 188         public function __construct( $conf = array() ) {
 189                 $this->mConf = $conf;
 190                 $this->mUrlProtocols = wfUrlProtocols();
 191                 $this->mExtLinkBracketedRegex = '/\[((' . wfUrlProtocols() . ')'.
 192                         self::EXT_LINK_URL_CLASS.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 193                 if ( isset( $conf['preprocessorClass'] ) ) {
 194                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 195                 } elseif ( defined( 'MW_COMPILED' ) ) {
 196                         # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
 197                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 198                 } elseif ( extension_loaded( 'domxml' ) ) {
 199                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 200                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 201                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 202                 } elseif ( extension_loaded( 'dom' ) ) {
 203                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 204                 } else {
 205                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 206                 }
 207                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 208         }
 209
 210         /**
 211          * Reduce memory usage to reduce the impact of circular references
 212          */
 213         function __destruct() {
 214                 if ( isset( $this->mLinkHolders ) ) {
 215                         unset( $this->mLinkHolders );
 216                 }
 217                 foreach ( $this as $name => $value ) {
 218                         unset( $this->$name );
 219                 }
 220         }
 221
 222         /**
 223          * Do various kinds of initialisation on the first call of the parser
 224          */
 225         function firstCallInit() {
 226                 if ( !$this->mFirstCall ) {
 227                         return;
 228                 }
 229                 $this->mFirstCall = false;
 230
 231                 wfProfileIn( __METHOD__ );
 232
 233                 CoreParserFunctions::register( $this );
 234                 CoreTagHooks::register( $this );
 235                 $this->initialiseVariables();
 236
 237                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 238                 wfProfileOut( __METHOD__ );
 239         }
 240
 241         /**
 242          * Clear Parser state
 243          *
 244          * @private
 245          */
 246         function clearState() {
 247                 wfProfileIn( __METHOD__ );
 248                 if ( $this->mFirstCall ) {
 249                         $this->firstCallInit();
 250                 }
 251                 $this->mOutput = new ParserOutput;
 252                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 253                 $this->mAutonumber = 0;
 254                 $this->mLastSection = '';
 255                 $this->mDTopen = false;
 256                 $this->mIncludeCount = array();
 257                 $this->mArgStack = false;
 258                 $this->mInPre = false;
 259                 $this->mLinkHolders = new LinkHolderArray( $this );
 260                 $this->mLinkID = 0;
 261                 $this->mRevisionObject = $this->mRevisionTimestamp =
 262                         $this->mRevisionId = $this->mRevisionUser = null;
 263                 $this->mVarCache = array();
 264                 $this->mUser = null;
 265
 266                 /**
 267                  * Prefix for temporary replacement strings for the multipass parser.
 268                  * \x07 should never appear in input as it's disallowed in XML.
 269                  * Using it at the front also gives us a little extra robustness
 270                  * since it shouldn't match when butted up against identifier-like
 271                  * string constructs.
 272                  *
 273                  * Must not consist of all title characters, or else it will change
 274                  * the behaviour of <nowiki> in a link.
 275                  */
 276                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 277                 # Changed to \x7f to allow XML double-parsing -- TS
 278                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 279                 $this->mStripState = new StripState( $this->mUniqPrefix );
 280
 281
 282                 # Clear these on every parse, bug 4549
 283                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 284
 285                 $this->mShowToc = true;
 286                 $this->mForceTocPosition = false;
 287                 $this->mIncludeSizes = array(
 288                         'post-expand' => 0,
 289                         'arg' => 0,
 290                 );
 291                 $this->mPPNodeCount = 0;
 292                 $this->mDefaultSort = false;
 293                 $this->mHeadings = array();
 294                 $this->mDoubleUnderscores = array();
 295                 $this->mExpensiveFunctionCount = 0;
 296
 297                 # Fix cloning
 298                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 299                         $this->mPreprocessor = null;
 300                 }
 301
 302                 wfRunHooks( 'ParserClearState', array( &$this ) );
 303                 wfProfileOut( __METHOD__ );
 304         }
 305
 306         /**
 307          * Convert wikitext to HTML
 308          * Do not call this function recursively.
 309          *
 310          * @param $text String: text we want to parse
 311          * @param $title Title object
 312          * @param $options ParserOptions
 313          * @param $linestart boolean
 314          * @param $clearState boolean
 315          * @param $revid Int: number to pass in {{REVISIONID}}
 316          * @return ParserOutput a ParserOutput
 317          */
 318         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 319                 /**
 320                  * First pass--just handle <nowiki> sections, pass the rest off
 321                  * to internalParse() which does all the real work.
 322                  */
 323
 324                 global $wgUseTidy, $wgAlwaysUseTidy, $wgDisableLangConversion, $wgDisableTitleConversion;
 325                 $fname = __METHOD__.'-' . wfGetCaller();
 326                 wfProfileIn( __METHOD__ );
 327                 wfProfileIn( $fname );
 328
 329                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 330
 331                 $oldRevisionId = $this->mRevisionId;
 332                 $oldRevisionObject = $this->mRevisionObject;
 333                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 334                 $oldRevisionUser = $this->mRevisionUser;
 335                 if ( $revid !== null ) {
 336                         $this->mRevisionId = $revid;
 337                         $this->mRevisionObject = null;
 338                         $this->mRevisionTimestamp = null;
 339                         $this->mRevisionUser = null;
 340                 }
 341
 342                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 343                 # No more strip!
 344                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 345                 $text = $this->internalParse( $text );
 346
 347                 $text = $this->mStripState->unstripGeneral( $text );
 348
 349                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 350                 $fixtags = array(
 351                         # french spaces, last one Guillemet-left
 352                         # only if there is something before the space
 353                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
 354                         # french spaces, Guillemet-right
 355                         '/(\\302\\253) /' => '\\1&#160;',
 356                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 357                 );
 358                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 359
 360                 $text = $this->doBlockLevels( $text, $linestart );
 361
 362                 $this->replaceLinkHolders( $text );
 363
 364                 /**
 365                  * The input doesn't get language converted if
 366                  * a) It's disabled
 367                  * b) Content isn't converted
 368                  * c) It's a conversion table
 369                  * d) it is an interface message (which is in the user language)
 370                  */
 371                 if ( !( $wgDisableLangConversion
 372                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 373                                 || $this->mTitle->isConversionTable()
 374                                 || $this->mOptions->getInterfaceMessage() ) ) {
 375
 376                         # The position of the convert() call should not be changed. it
 377                         # assumes that the links are all replaced and the only thing left
 378                         # is the <nowiki> mark.
 379                         $text = $this->getFunctionLang()->convert( $text );
 380                 }
 381
 382                 /**
 383                  * A converted title will be provided in the output object if title and
 384                  * content conversion are enabled, the article text does not contain
 385                  * a conversion-suppressing double-underscore tag, and no
 386                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 387                  * automatic link conversion.
 388                  */
 389                 if ( !( $wgDisableLangConversion
 390                                 || $wgDisableTitleConversion
 391                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 392                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 393                                 || $this->mOutput->getDisplayTitle() !== false ) )
 394                 {
 395                         $convruletitle = $this->getFunctionLang()->getConvRuleTitle();
 396                         if ( $convruletitle ) {
 397                                 $this->mOutput->setTitleText( $convruletitle );
 398                         } else {
 399                                 $titleText = $this->getFunctionLang()->convertTitle( $title );
 400                                 $this->mOutput->setTitleText( $titleText );
 401                         }
 402                 }
 403
 404                 $text = $this->mStripState->unstripNoWiki( $text );
 405
 406                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 407
 408                 $text = $this->replaceTransparentTags( $text );
 409                 $text = $this->mStripState->unstripGeneral( $text );
 410
 411                 $text = Sanitizer::normalizeCharReferences( $text );
 412
 413                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 414                         $text = MWTidy::tidy( $text );
 415                 } else {
 416                         # attempt to sanitize at least some nesting problems
 417                         # (bug #2702 and quite a few others)
 418                         $tidyregs = array(
 419                                 # ''Something [http://www.cool.com cool''] -->
 420                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 421                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 422                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 423                                 # fix up an anchor inside another anchor, only
 424                                 # at least for a single single nested link (bug 3695)
 425                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 426                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 427                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 428                                 # contains a div, so fix it up here; replace
 429                                 # div with escaped text
 430                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 431                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 432                                 # remove empty italic or bold tag pairs, some
 433                                 # introduced by rules above
 434                                 '/<([bi])><\/\\1>/' => '',
 435                         );
 436
 437                         $text = preg_replace(
 438                                 array_keys( $tidyregs ),
 439                                 array_values( $tidyregs ),
 440                                 $text );
 441                 }
 442                 global $wgExpensiveParserFunctionLimit;
 443                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 444                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 445                 }
 446
 447                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 448
 449                 # Information on include size limits, for the benefit of users who try to skirt them
 450                 if ( $this->mOptions->getEnableLimitReport() ) {
 451                         $max = $this->mOptions->getMaxIncludeSize();
 452                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 453                         $limitReport =
 454                                 "NewPP limit report\n" .
 455                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 456                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 457                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 458                                 $PFreport;
 459                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 460                         $text .= "\n<!-- \n$limitReport-->\n";
 461                 }
 462                 $this->mOutput->setText( $text );
 463
 464                 $this->mRevisionId = $oldRevisionId;
 465                 $this->mRevisionObject = $oldRevisionObject;
 466                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 467                 $this->mRevisionUser = $oldRevisionUser;
 468                 wfProfileOut( $fname );
 469                 wfProfileOut( __METHOD__ );
 470
 471                 return $this->mOutput;
 472         }
 473
 474         /**
 475          * Recursive parser entry point that can be called from an extension tag
 476          * hook.
 477          *
 478          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 479          *
 480          * @param $text String: text extension wants to have parsed
 481          * @param $frame PPFrame: The frame to use for expanding any template variables
 482          *
 483          * @return string
 484          */
 485         function recursiveTagParse( $text, $frame=false ) {
 486                 wfProfileIn( __METHOD__ );
 487                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 488                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 $text = $this->internalParse( $text, false, $frame );
 490                 wfProfileOut( __METHOD__ );
 491                 return $text;
 492         }
 493
 494         /**
 495          * Expand templates and variables in the text, producing valid, static wikitext.
 496          * Also removes comments.
 497          */
 498         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 499                 wfProfileIn( __METHOD__ );
 500                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 501                 if ( $revid !== null ) {
 502                         $this->mRevisionId = $revid;
 503                 }
 504                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 505                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 506                 $text = $this->replaceVariables( $text );
 507                 $text = $this->mStripState->unstripBoth( $text );
 508                 wfProfileOut( __METHOD__ );
 509                 return $text;
 510         }
 511
 512         /**
 513          * Recursive parser entry point that can be called from an extension tag
 514          * hook.
 515          *
 516          * @param $text String: text to be expanded
 517          * @param $frame PPFrame: The frame to use for expanding any template variables
 518          * @return String
 519          */
 520         public function recursivePreprocess( $text, $frame = false ) {
 521                 wfProfileIn( __METHOD__ );
 522                 $text = $this->replaceVariables( $text, $frame );
 523                 $text = $this->mStripState->unstripBoth( $text );
 524                 wfProfileOut( __METHOD__ );
 525                 return $text;
 526         }
 527
 528         /**
 529          * Process the wikitext for the ?preload= feature. (bug 5210)
 530          *
 531          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 532          * comments, templates, arguments, tags hooks and parser functions are untouched.
 533          *
 534          * @param $text String
 535          * @param $title Title
 536          * @param $options ParserOptions
 537          * @return String
 538          */
 539         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 540                 # Parser (re)initialisation
 541                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 542
 543                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 544                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 545                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 546                 $text = $this->mStripState->unstripBoth( $text );
 547                 return $text;
 548         }
 549
 550         /**
 551          * Get a random string
 552          *
 553          * @return string
 554          */
 555         static public function getRandomString() {
 556                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 557         }
 558
 559         /**
 560          * Set the current user.
 561          * Should only be used when doing pre-save transform.
 562          *
 563          * @param $user Mixed: User object or null (to reset)
 564          */
 565         function setUser( $user ) {
 566                 $this->mUser = $user;
 567         }
 568
 569         /**
 570          * Accessor for mUniqPrefix.
 571          *
 572          * @return String
 573          */
 574         public function uniqPrefix() {
 575                 if ( !isset( $this->mUniqPrefix ) ) {
 576                         # @todo FIXME: This is probably *horribly wrong*
 577                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 578                         # if this is called for a parser cache hit, the parser may not
 579                         # have ever been initialized in the first place.
 580                         # Not really sure what the heck is supposed to be going on here.
 581                         return '';
 582                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 583                 }
 584                 return $this->mUniqPrefix;
 585         }
 586
 587         /**
 588          * Set the context title
 589          *
 590          * @param $t Title
 591          */
 592         function setTitle( $t ) {
 593                 if ( !$t || $t instanceof FakeTitle ) {
 594                         $t = Title::newFromText( 'NO TITLE' );
 595                 }
 596
 597                 if ( strval( $t->getFragment() ) !== '' ) {
 598                         # Strip the fragment to avoid various odd effects
 599                         $this->mTitle = clone $t;
 600                         $this->mTitle->setFragment( '' );
 601                 } else {
 602                         $this->mTitle = $t;
 603                 }
 604         }
 605
 606         /**
 607          * Accessor for the Title object
 608          *
 609          * @return Title object
 610          */
 611         function getTitle() {
 612                 return $this->mTitle;
 613         }
 614
 615         /**
 616          * Accessor/mutator for the Title object
 617          *
 618          * @param $x New Title object or null to just get the current one
 619          * @return Title object
 620          */
 621         function Title( $x = null ) {
 622                 return wfSetVar( $this->mTitle, $x );
 623         }
 624
 625         /**
 626          * Set the output type
 627          *
 628          * @param $ot Integer: new value
 629          */
 630         function setOutputType( $ot ) {
 631                 $this->mOutputType = $ot;
 632                 # Shortcut alias
 633                 $this->ot = array(
 634                         'html' => $ot == self::OT_HTML,
 635                         'wiki' => $ot == self::OT_WIKI,
 636                         'pre' => $ot == self::OT_PREPROCESS,
 637                         'plain' => $ot == self::OT_PLAIN,
 638                 );
 639         }
 640
 641         /**
 642          * Accessor/mutator for the output type
 643          *
 644          * @param $x New value or null to just get the current one
 645          * @return Integer
 646          */
 647         function OutputType( $x = null ) {
 648                 return wfSetVar( $this->mOutputType, $x );
 649         }
 650
 651         /**
 652          * Get the ParserOutput object
 653          *
 654          * @return ParserOutput object
 655          */
 656         function getOutput() {
 657                 return $this->mOutput;
 658         }
 659
 660         /**
 661          * Get the ParserOptions object
 662          *
 663          * @return ParserOptions object
 664          */
 665         function getOptions() {
 666                 return $this->mOptions;
 667         }
 668
 669         /**
 670          * Accessor/mutator for the ParserOptions object
 671          *
 672          * @param $x New value or null to just get the current one
 673          * @return Current ParserOptions object
 674          */
 675         function Options( $x = null ) {
 676                 return wfSetVar( $this->mOptions, $x );
 677         }
 678
 679         /**
 680          * @return int
 681          */
 682         function nextLinkID() {
 683                 return $this->mLinkID++;
 684         }
 685
 686         /**
 687          * @param $id int
 688          */
 689         function setLinkID( $id ) {
 690                 $this->mLinkID = $id;
 691         }
 692
 693         /**
 694          * @return Language
 695          */
 696         function getFunctionLang() {
 697                 $target = $this->mOptions->getTargetLanguage();
 698                 if ( $target !== null ) {
 699                         return $target;
 700                 } elseif( $this->mOptions->getInterfaceMessage() ) {
 701                         global $wgLang;
 702                         return $wgLang;
 703                 } elseif( is_null( $this->mTitle ) ) {
 704                         throw new MWException( __METHOD__.': $this->mTitle is null' );
 705                 }
 706                 return $this->mTitle->getPageLanguage();
 707         }
 708
 709         /**
 710          * Get a User object either from $this->mUser, if set, or from the
 711          * ParserOptions object otherwise
 712          *
 713          * @return User object
 714          */
 715         function getUser() {
 716                 if ( !is_null( $this->mUser ) ) {
 717                         return $this->mUser;
 718                 }
 719                 return $this->mOptions->getUser();
 720         }
 721
 722         /**
 723          * Get a preprocessor object
 724          *
 725          * @return Preprocessor instance
 726          */
 727         function getPreprocessor() {
 728                 if ( !isset( $this->mPreprocessor ) ) {
 729                         $class = $this->mPreprocessorClass;
 730                         $this->mPreprocessor = new $class( $this );
 731                 }
 732                 return $this->mPreprocessor;
 733         }
 734
 735         /**
 736          * Replaces all occurrences of HTML-style comments and the given tags
 737          * in the text with a random marker and returns the next text. The output
 738          * parameter $matches will be an associative array filled with data in
 739          * the form:
 740          *   'UNIQ-xxxxx' => array(
 741          *     'element',
 742          *     'tag content',
 743          *     array( 'param' => 'x' ),
 744          *     '<element param="x">tag content</element>' ) )
 745          *
 746          * @param $elements array list of element names. Comments are always extracted.
 747          * @param $text string Source text string.
 748          * @param $matches array Out parameter, Array: extracted tags
 749          * @param $uniq_prefix string
 750          * @return String: stripped text
 751          */
 752         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 753                 static $n = 1;
 754                 $stripped = '';
 755                 $matches = array();
 756
 757                 $taglist = implode( '|', $elements );
 758                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 759
 760                 while ( $text != '' ) {
 761                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 762                         $stripped .= $p[0];
 763                         if ( count( $p ) < 5 ) {
 764                                 break;
 765                         }
 766                         if ( count( $p ) > 5 ) {
 767                                 # comment
 768                                 $element    = $p[4];
 769                                 $attributes = '';
 770                                 $close      = '';
 771                                 $inside     = $p[5];
 772                         } else {
 773                                 # tag
 774                                 $element    = $p[1];
 775                                 $attributes = $p[2];
 776                                 $close      = $p[3];
 777                                 $inside     = $p[4];
 778                         }
 779
 780                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 781                         $stripped .= $marker;
 782
 783                         if ( $close === '/>' ) {
 784                                 # Empty element tag, <tag />
 785                                 $content = null;
 786                                 $text = $inside;
 787                                 $tail = null;
 788                         } else {
 789                                 if ( $element === '!--' ) {
 790                                         $end = '/(-->)/';
 791                                 } else {
 792                                         $end = "/(<\\/$element\\s*>)/i";
 793                                 }
 794                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 795                                 $content = $q[0];
 796                                 if ( count( $q ) < 3 ) {
 797                                         # No end tag -- let it run out to the end of the text.
 798                                         $tail = '';
 799                                         $text = '';
 800                                 } else {
 801                                         $tail = $q[1];
 802                                         $text = $q[2];
 803                                 }
 804                         }
 805
 806                         $matches[$marker] = array( $element,
 807                                 $content,
 808                                 Sanitizer::decodeTagAttributes( $attributes ),
 809                                 "<$element$attributes$close$content$tail" );
 810                 }
 811                 return $stripped;
 812         }
 813
 814         /**
 815          * Get a list of strippable XML-like elements
 816          *
 817          * @return array
 818          */
 819         function getStripList() {
 820                 return $this->mStripList;
 821         }
 822
 823         /**
 824          * Add an item to the strip state
 825          * Returns the unique tag which must be inserted into the stripped text
 826          * The tag will be replaced with the original text in unstrip()
 827          *
 828          * @param $text string
 829          *
 830          * @return string
 831          */
 832         function insertStripItem( $text ) {
 833                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 834                 $this->mMarkerIndex++;
 835                 $this->mStripState->addGeneral( $rnd, $text );
 836                 return $rnd;
 837         }
 838
 839         /**
 840          * parse the wiki syntax used to render tables
 841          *
 842          * @private
 843          */
 844         function doTableStuff( $text ) {
 845                 wfProfileIn( __METHOD__ );
 846
 847                 $lines = StringUtils::explode( "\n", $text );
 848                 $out = '';
 849                 $td_history = array(); # Is currently a td tag open?
 850                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 851                 $tr_history = array(); # Is currently a tr tag open?
 852                 $tr_attributes = array(); # history of tr attributes
 853                 $has_opened_tr = array(); # Did this table open a <tr> element?
 854                 $indent_level = 0; # indent level of the table
 855
 856                 foreach ( $lines as $outLine ) {
 857                         $line = trim( $outLine );
 858
 859                         if ( $line === '' ) { # empty line, go to next line
 860                                 $out .= $outLine."\n";
 861                                 continue;
 862                         }
 863
 864                         $first_character = $line[0];
 865                         $matches = array();
 866
 867                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 868                                 # First check if we are starting a new table
 869                                 $indent_level = strlen( $matches[1] );
 870
 871                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 872                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 873
 874                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 875                                 array_push( $td_history , false );
 876                                 array_push( $last_tag_history , '' );
 877                                 array_push( $tr_history , false );
 878                                 array_push( $tr_attributes , '' );
 879                                 array_push( $has_opened_tr , false );
 880                         } elseif ( count( $td_history ) == 0 ) {
 881                                 # Don't do any of the following
 882                                 $out .= $outLine."\n";
 883                                 continue;
 884                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 885                                 # We are ending a table
 886                                 $line = '</table>' . substr( $line , 2 );
 887                                 $last_tag = array_pop( $last_tag_history );
 888
 889                                 if ( !array_pop( $has_opened_tr ) ) {
 890                                         $line = "<tr><td></td></tr>{$line}";
 891                                 }
 892
 893                                 if ( array_pop( $tr_history ) ) {
 894                                         $line = "</tr>{$line}";
 895                                 }
 896
 897                                 if ( array_pop( $td_history ) ) {
 898                                         $line = "</{$last_tag}>{$line}";
 899                                 }
 900                                 array_pop( $tr_attributes );
 901                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 902                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 903                                 # Now we have a table row
 904                                 $line = preg_replace( '#^\|-+#', '', $line );
 905
 906                                 # Whats after the tag is now only attributes
 907                                 $attributes = $this->mStripState->unstripBoth( $line );
 908                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 909                                 array_pop( $tr_attributes );
 910                                 array_push( $tr_attributes, $attributes );
 911
 912                                 $line = '';
 913                                 $last_tag = array_pop( $last_tag_history );
 914                                 array_pop( $has_opened_tr );
 915                                 array_push( $has_opened_tr , true );
 916
 917                                 if ( array_pop( $tr_history ) ) {
 918                                         $line = '</tr>';
 919                                 }
 920
 921                                 if ( array_pop( $td_history ) ) {
 922                                         $line = "</{$last_tag}>{$line}";
 923                                 }
 924
 925                                 $outLine = $line;
 926                                 array_push( $tr_history , false );
 927                                 array_push( $td_history , false );
 928                                 array_push( $last_tag_history , '' );
 929                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 930                                 # This might be cell elements, td, th or captions
 931                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 932                                         $first_character = '+';
 933                                         $line = substr( $line , 1 );
 934                                 }
 935
 936                                 $line = substr( $line , 1 );
 937
 938                                 if ( $first_character === '!' ) {
 939                                         $line = str_replace( '!!' , '||' , $line );
 940                                 }
 941
 942                                 # Split up multiple cells on the same line.
 943                                 # FIXME : This can result in improper nesting of tags processed
 944                                 # by earlier parser steps, but should avoid splitting up eg
 945                                 # attribute values containing literal "||".
 946                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 947
 948                                 $outLine = '';
 949
 950                                 # Loop through each table cell
 951                                 foreach ( $cells as $cell ) {
 952                                         $previous = '';
 953                                         if ( $first_character !== '+' ) {
 954                                                 $tr_after = array_pop( $tr_attributes );
 955                                                 if ( !array_pop( $tr_history ) ) {
 956                                                         $previous = "<tr{$tr_after}>\n";
 957                                                 }
 958                                                 array_push( $tr_history , true );
 959                                                 array_push( $tr_attributes , '' );
 960                                                 array_pop( $has_opened_tr );
 961                                                 array_push( $has_opened_tr , true );
 962                                         }
 963
 964                                         $last_tag = array_pop( $last_tag_history );
 965
 966                                         if ( array_pop( $td_history ) ) {
 967                                                 $previous = "</{$last_tag}>\n{$previous}";
 968                                         }
 969
 970                                         if ( $first_character === '|' ) {
 971                                                 $last_tag = 'td';
 972                                         } elseif ( $first_character === '!' ) {
 973                                                 $last_tag = 'th';
 974                                         } elseif ( $first_character === '+' ) {
 975                                                 $last_tag = 'caption';
 976                                         } else {
 977                                                 $last_tag = '';
 978                                         }
 979
 980                                         array_push( $last_tag_history , $last_tag );
 981
 982                                         # A cell could contain both parameters and data
 983                                         $cell_data = explode( '|' , $cell , 2 );
 984
 985                                         # Bug 553: Note that a '|' inside an invalid link should not
 986                                         # be mistaken as delimiting cell parameters
 987                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 988                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 989                                         } elseif ( count( $cell_data ) == 1 ) {
 990                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 991                                         } else {
 992                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 993                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 994                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 995                                         }
 996
 997                                         $outLine .= $cell;
 998                                         array_push( $td_history , true );
 999                                 }
1000                         }
1001                         $out .= $outLine . "\n";
1002                 }
1003
1004                 # Closing open td, tr && table
1005                 while ( count( $td_history ) > 0 ) {
1006                         if ( array_pop( $td_history ) ) {
1007                                 $out .= "</td>\n";
1008                         }
1009                         if ( array_pop( $tr_history ) ) {
1010                                 $out .= "</tr>\n";
1011                         }
1012                         if ( !array_pop( $has_opened_tr ) ) {
1013                                 $out .= "<tr><td></td></tr>\n" ;
1014                         }
1015
1016                         $out .= "</table>\n";
1017                 }
1018
1019                 # Remove trailing line-ending (b/c)
1020                 if ( substr( $out, -1 ) === "\n" ) {
1021                         $out = substr( $out, 0, -1 );
1022                 }
1023
1024                 # special case: don't return empty table
1025                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1026                         $out = '';
1027                 }
1028
1029                 wfProfileOut( __METHOD__ );
1030
1031                 return $out;
1032         }
1033
1034         /**
1035          * Helper function for parse() that transforms wiki markup into
1036          * HTML. Only called for $mOutputType == self::OT_HTML.
1037          *
1038          * @private
1039          *
1040          * @param $text string
1041          * @param $isMain bool
1042          * @param $frame bool
1043          *
1044          * @return string
1045          */
1046         function internalParse( $text, $isMain = true, $frame = false ) {
1047                 wfProfileIn( __METHOD__ );
1048
1049                 $origText = $text;
1050
1051                 # Hook to suspend the parser in this state
1052                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1053                         wfProfileOut( __METHOD__ );
1054                         return $text ;
1055                 }
1056
1057                 # if $frame is provided, then use $frame for replacing any variables
1058                 if ( $frame ) {
1059                         # use frame depth to infer how include/noinclude tags should be handled
1060                         # depth=0 means this is the top-level document; otherwise it's an included document
1061                         if ( !$frame->depth ) {
1062                                 $flag = 0;
1063                         } else {
1064                                 $flag = Parser::PTD_FOR_INCLUSION;
1065                         }
1066                         $dom = $this->preprocessToDom( $text, $flag );
1067                         $text = $frame->expand( $dom );
1068                 } else {
1069                         # if $frame is not provided, then use old-style replaceVariables
1070                         $text = $this->replaceVariables( $text );
1071                 }
1072
1073                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1074                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1075
1076                 # Tables need to come after variable replacement for things to work
1077                 # properly; putting them before other transformations should keep
1078                 # exciting things like link expansions from showing up in surprising
1079                 # places.
1080                 $text = $this->doTableStuff( $text );
1081
1082                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1083
1084                 $text = $this->doDoubleUnderscore( $text );
1085
1086                 $text = $this->doHeadings( $text );
1087                 if ( $this->mOptions->getUseDynamicDates() ) {
1088                         $df = DateFormatter::getInstance();
1089                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1090                 }
1091                 $text = $this->replaceInternalLinks( $text );
1092                 $text = $this->doAllQuotes( $text );
1093                 $text = $this->replaceExternalLinks( $text );
1094
1095                 # replaceInternalLinks may sometimes leave behind
1096                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1097                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1098
1099                 $text = $this->doMagicLinks( $text );
1100                 $text = $this->formatHeadings( $text, $origText, $isMain );
1101
1102                 wfProfileOut( __METHOD__ );
1103                 return $text;
1104         }
1105
1106         /**
1107          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1108          * magic external links.
1109          *
1110          * DML
1111          * @private
1112          *
1113          * @param $text string
1114          *
1115          * @return string
1116          */
1117         function doMagicLinks( $text ) {
1118                 wfProfileIn( __METHOD__ );
1119                 $prots = wfUrlProtocolsWithoutProtRel();
1120                 $urlChar = self::EXT_LINK_URL_CLASS;
1121                 $text = preg_replace_callback(
1122                         '!(?:                           # Start cases
1123                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1124                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1125                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1126                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1127                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1128                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1129                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1130                                         [0-9Xx]                 # check digit
1131                                         \b)
1132                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1133                 wfProfileOut( __METHOD__ );
1134                 return $text;
1135         }
1136
1137         /**
1138          * @throws MWException
1139          * @param $m array
1140          * @return HTML|string
1141          */
1142         function magicLinkCallback( $m ) {
1143                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1144                         # Skip anchor
1145                         return $m[0];
1146                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1147                         # Skip HTML element
1148                         return $m[0];
1149                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1150                         # Free external link
1151                         return $this->makeFreeExternalLink( $m[0] );
1152                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1153                         # RFC or PMID
1154                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1155                                 $keyword = 'RFC';
1156                                 $urlmsg = 'rfcurl';
1157                                 $CssClass = 'mw-magiclink-rfc';
1158                                 $id = $m[4];
1159                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1160                                 $keyword = 'PMID';
1161                                 $urlmsg = 'pubmedurl';
1162                                 $CssClass = 'mw-magiclink-pmid';
1163                                 $id = $m[4];
1164                         } else {
1165                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1166                                         substr( $m[0], 0, 20 ) . '"' );
1167                         }
1168                         $url = wfMsgForContent( $urlmsg, $id );
1169                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1170                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1171                         # ISBN
1172                         $isbn = $m[5];
1173                         $num = strtr( $isbn, array(
1174                                 '-' => '',
1175                                 ' ' => '',
1176                                 'x' => 'X',
1177                         ));
1178                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1179                         return'<a href="' .
1180                                 $titleObj->escapeLocalUrl() .
1181                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1182                 } else {
1183                         return $m[0];
1184                 }
1185         }
1186
1187         /**
1188          * Make a free external link, given a user-supplied URL
1189          *
1190          * @param $url string
1191          *
1192          * @return string HTML
1193          * @private
1194          */
1195         function makeFreeExternalLink( $url ) {
1196                 wfProfileIn( __METHOD__ );
1197
1198                 $trail = '';
1199
1200                 # The characters '<' and '>' (which were escaped by
1201                 # removeHTMLtags()) should not be included in
1202                 # URLs, per RFC 2396.
1203                 $m2 = array();
1204                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1205                         $trail = substr( $url, $m2[0][1] ) . $trail;
1206                         $url = substr( $url, 0, $m2[0][1] );
1207                 }
1208
1209                 # Move trailing punctuation to $trail
1210                 $sep = ',;\.:!?';
1211                 # If there is no left bracket, then consider right brackets fair game too
1212                 if ( strpos( $url, '(' ) === false ) {
1213                         $sep .= ')';
1214                 }
1215
1216                 $numSepChars = strspn( strrev( $url ), $sep );
1217                 if ( $numSepChars ) {
1218                         $trail = substr( $url, -$numSepChars ) . $trail;
1219                         $url = substr( $url, 0, -$numSepChars );
1220                 }
1221
1222                 $url = Sanitizer::cleanUrl( $url );
1223
1224                 # Is this an external image?
1225                 $text = $this->maybeMakeExternalImage( $url );
1226                 if ( $text === false ) {
1227                         # Not an image, make a link
1228                         $text = Linker::makeExternalLink( $url, $this->getFunctionLang()->markNoConversion($url), true, 'free',
1229                                 $this->getExternalLinkAttribs( $url ) );
1230                         # Register it in the output object...
1231                         # Replace unnecessary URL escape codes with their equivalent characters
1232                         $pasteurized = self::replaceUnusualEscapes( $url );
1233                         $this->mOutput->addExternalLink( $pasteurized );
1234                 }
1235                 wfProfileOut( __METHOD__ );
1236                 return $text . $trail;
1237         }
1238
1239
1240         /**
1241          * Parse headers and return html
1242          *
1243          * @private
1244          *
1245          * @param $text string
1246          *
1247          * @return string
1248          */
1249         function doHeadings( $text ) {
1250                 wfProfileIn( __METHOD__ );
1251                 for ( $i = 6; $i >= 1; --$i ) {
1252                         $h = str_repeat( '=', $i );
1253                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1254                           "<h$i>\\1</h$i>", $text );
1255                 }
1256                 wfProfileOut( __METHOD__ );
1257                 return $text;
1258         }
1259
1260         /**
1261          * Replace single quotes with HTML markup
1262          * @private
1263          *
1264          * @param $text string
1265          *
1266          * @return string the altered text
1267          */
1268         function doAllQuotes( $text ) {
1269                 wfProfileIn( __METHOD__ );
1270                 $outtext = '';
1271                 $lines = StringUtils::explode( "\n", $text );
1272                 foreach ( $lines as $line ) {
1273                         $outtext .= $this->doQuotes( $line ) . "\n";
1274                 }
1275                 $outtext = substr( $outtext, 0,-1 );
1276                 wfProfileOut( __METHOD__ );
1277                 return $outtext;
1278         }
1279
1280         /**
1281          * Helper function for doAllQuotes()
1282          *
1283          * @param $text string
1284          *
1285          * @return string
1286          */
1287         public function doQuotes( $text ) {
1288                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1289                 if ( count( $arr ) == 1 ) {
1290                         return $text;
1291                 } else {
1292                         # First, do some preliminary work. This may shift some apostrophes from
1293                         # being mark-up to being text. It also counts the number of occurrences
1294                         # of bold and italics mark-ups.
1295                         $numbold = 0;
1296                         $numitalics = 0;
1297                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1298                                 if ( ( $i % 2 ) == 1 ) {
1299                                         # If there are ever four apostrophes, assume the first is supposed to
1300                                         # be text, and the remaining three constitute mark-up for bold text.
1301                                         if ( strlen( $arr[$i] ) == 4 ) {
1302                                                 $arr[$i-1] .= "'";
1303                                                 $arr[$i] = "'''";
1304                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1305                                                 # If there are more than 5 apostrophes in a row, assume they're all
1306                                                 # text except for the last 5.
1307                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1308                                                 $arr[$i] = "'''''";
1309                                         }
1310                                         # Count the number of occurrences of bold and italics mark-ups.
1311                                         # We are not counting sequences of five apostrophes.
1312                                         if ( strlen( $arr[$i] ) == 2 ) {
1313                                                 $numitalics++;
1314                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1315                                                 $numbold++;
1316                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1317                                                 $numitalics++;
1318                                                 $numbold++;
1319                                         }
1320                                 }
1321                         }
1322
1323                         # If there is an odd number of both bold and italics, it is likely
1324                         # that one of the bold ones was meant to be an apostrophe followed
1325                         # by italics. Which one we cannot know for certain, but it is more
1326                         # likely to be one that has a single-letter word before it.
1327                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1328                                 $i = 0;
1329                                 $firstsingleletterword = -1;
1330                                 $firstmultiletterword = -1;
1331                                 $firstspace = -1;
1332                                 foreach ( $arr as $r ) {
1333                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1334                                                 $x1 = substr( $arr[$i-1], -1 );
1335                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1336                                                 if ( $x1 === ' ' ) {
1337                                                         if ( $firstspace == -1 ) {
1338                                                                 $firstspace = $i;
1339                                                         }
1340                                                 } elseif ( $x2 === ' ') {
1341                                                         if ( $firstsingleletterword == -1 ) {
1342                                                                 $firstsingleletterword = $i;
1343                                                         }
1344                                                 } else {
1345                                                         if ( $firstmultiletterword == -1 ) {
1346                                                                 $firstmultiletterword = $i;
1347                                                         }
1348                                                 }
1349                                         }
1350                                         $i++;
1351                                 }
1352
1353                                 # If there is a single-letter word, use it!
1354                                 if ( $firstsingleletterword > -1 ) {
1355                                         $arr[$firstsingleletterword] = "''";
1356                                         $arr[$firstsingleletterword-1] .= "'";
1357                                 } elseif ( $firstmultiletterword > -1 ) {
1358                                         # If not, but there's a multi-letter word, use that one.
1359                                         $arr[$firstmultiletterword] = "''";
1360                                         $arr[$firstmultiletterword-1] .= "'";
1361                                 } elseif ( $firstspace > -1 ) {
1362                                         # ... otherwise use the first one that has neither.
1363                                         # (notice that it is possible for all three to be -1 if, for example,
1364                                         # there is only one pentuple-apostrophe in the line)
1365                                         $arr[$firstspace] = "''";
1366                                         $arr[$firstspace-1] .= "'";
1367                                 }
1368                         }
1369
1370                         # Now let's actually convert our apostrophic mush to HTML!
1371                         $output = '';
1372                         $buffer = '';
1373                         $state = '';
1374                         $i = 0;
1375                         foreach ( $arr as $r ) {
1376                                 if ( ( $i % 2 ) == 0 ) {
1377                                         if ( $state === 'both' ) {
1378                                                 $buffer .= $r;
1379                                         } else {
1380                                                 $output .= $r;
1381                                         }
1382                                 } else {
1383                                         if ( strlen( $r ) == 2 ) {
1384                                                 if ( $state === 'i' ) {
1385                                                         $output .= '</i>'; $state = '';
1386                                                 } elseif ( $state === 'bi' ) {
1387                                                         $output .= '</i>'; $state = 'b';
1388                                                 } elseif ( $state === 'ib' ) {
1389                                                         $output .= '</b></i><b>'; $state = 'b';
1390                                                 } elseif ( $state === 'both' ) {
1391                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1392                                                 } else { # $state can be 'b' or ''
1393                                                         $output .= '<i>'; $state .= 'i';
1394                                                 }
1395                                         } elseif ( strlen( $r ) == 3 ) {
1396                                                 if ( $state === 'b' ) {
1397                                                         $output .= '</b>'; $state = '';
1398                                                 } elseif ( $state === 'bi' ) {
1399                                                         $output .= '</i></b><i>'; $state = 'i';
1400                                                 } elseif ( $state === 'ib' ) {
1401                                                         $output .= '</b>'; $state = 'i';
1402                                                 } elseif ( $state === 'both' ) {
1403                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1404                                                 } else { # $state can be 'i' or ''
1405                                                         $output .= '<b>'; $state .= 'b';
1406                                                 }
1407                                         } elseif ( strlen( $r ) == 5 ) {
1408                                                 if ( $state === 'b' ) {
1409                                                         $output .= '</b><i>'; $state = 'i';
1410                                                 } elseif ( $state === 'i' ) {
1411                                                         $output .= '</i><b>'; $state = 'b';
1412                                                 } elseif ( $state === 'bi' ) {
1413                                                         $output .= '</i></b>'; $state = '';
1414                                                 } elseif ( $state === 'ib' ) {
1415                                                         $output .= '</b></i>'; $state = '';
1416                                                 } elseif ( $state === 'both' ) {
1417                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1418                                                 } else { # ($state == '')
1419                                                         $buffer = ''; $state = 'both';
1420                                                 }
1421                                         }
1422                                 }
1423                                 $i++;
1424                         }
1425                         # Now close all remaining tags.  Notice that the order is important.
1426                         if ( $state === 'b' || $state === 'ib' ) {
1427                                 $output .= '</b>';
1428                         }
1429                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1430                                 $output .= '</i>';
1431                         }
1432                         if ( $state === 'bi' ) {
1433                                 $output .= '</b>';
1434                         }
1435                         # There might be lonely ''''', so make sure we have a buffer
1436                         if ( $state === 'both' && $buffer ) {
1437                                 $output .= '<b><i>'.$buffer.'</i></b>';
1438                         }
1439                         return $output;
1440                 }
1441         }
1442
1443         /**
1444          * Replace external links (REL)
1445          *
1446          * Note: this is all very hackish and the order of execution matters a lot.
1447          * Make sure to run maintenance/parserTests.php if you change this code.
1448          *
1449          * @private
1450          *
1451          * @param $text string
1452          *
1453          * @return string
1454          */
1455         function replaceExternalLinks( $text ) {
1456                 wfProfileIn( __METHOD__ );
1457
1458                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1459                 $s = array_shift( $bits );
1460
1461                 $i = 0;
1462                 while ( $i<count( $bits ) ) {
1463                         $url = $bits[$i++];
1464                         $protocol = $bits[$i++];
1465                         $text = $bits[$i++];
1466                         $trail = $bits[$i++];
1467
1468                         # The characters '<' and '>' (which were escaped by
1469                         # removeHTMLtags()) should not be included in
1470                         # URLs, per RFC 2396.
1471                         $m2 = array();
1472                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1473                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1474                                 $url = substr( $url, 0, $m2[0][1] );
1475                         }
1476
1477                         # If the link text is an image URL, replace it with an <img> tag
1478                         # This happened by accident in the original parser, but some people used it extensively
1479                         $img = $this->maybeMakeExternalImage( $text );
1480                         if ( $img !== false ) {
1481                                 $text = $img;
1482                         }
1483
1484                         $dtrail = '';
1485
1486                         # Set linktype for CSS - if URL==text, link is essentially free
1487                         $linktype = ( $text === $url ) ? 'free' : 'text';
1488
1489                         # No link text, e.g. [http://domain.tld/some.link]
1490                         if ( $text == '' ) {
1491                                 # Autonumber
1492                                 $langObj = $this->getFunctionLang();
1493                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1494                                 $linktype = 'autonumber';
1495                         } else {
1496                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1497                                 # Check for trail
1498                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1499                         }
1500
1501                         $text = $this->getFunctionLang()->markNoConversion( $text );
1502
1503                         $url = Sanitizer::cleanUrl( $url );
1504
1505                         # Use the encoded URL
1506                         # This means that users can paste URLs directly into the text
1507                         # Funny characters like ö aren't valid in URLs anyway
1508                         # This was changed in August 2004
1509                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1510                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1511
1512                         # Register link in the output object.
1513                         # Replace unnecessary URL escape codes with the referenced character
1514                         # This prevents spammers from hiding links from the filters
1515                         $pasteurized = self::replaceUnusualEscapes( $url );
1516                         $this->mOutput->addExternalLink( $pasteurized );
1517                 }
1518
1519                 wfProfileOut( __METHOD__ );
1520                 return $s;
1521         }
1522
1523         /**
1524          * Get an associative array of additional HTML attributes appropriate for a
1525          * particular external link.  This currently may include rel => nofollow
1526          * (depending on configuration, namespace, and the URL's domain) and/or a
1527          * target attribute (depending on configuration).
1528          *
1529          * @param $url String|bool optional URL, to extract the domain from for rel =>
1530          *   nofollow if appropriate
1531          * @return Array associative array of HTML attributes
1532          */
1533         function getExternalLinkAttribs( $url = false ) {
1534                 $attribs = array();
1535                 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1536                 $ns = $this->mTitle->getNamespace();
1537                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1538                                 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1539                 {
1540                         $attribs['rel'] = 'nofollow';
1541                 }
1542                 if ( $this->mOptions->getExternalLinkTarget() ) {
1543                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1544                 }
1545                 return $attribs;
1546         }
1547
1548         /**
1549          * Replace unusual URL escape codes with their equivalent characters
1550          *
1551          * @param $url String
1552          * @return String
1553          *
1554          * @todo  This can merge genuinely required bits in the path or query string,
1555          *        breaking legit URLs. A proper fix would treat the various parts of
1556          *        the URL differently; as a workaround, just use the output for
1557          *        statistical records, not for actual linking/output.
1558          */
1559         static function replaceUnusualEscapes( $url ) {
1560                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1561                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1562         }
1563
1564         /**
1565          * Callback function used in replaceUnusualEscapes().
1566          * Replaces unusual URL escape codes with their equivalent character
1567          *
1568          * @param $matches array
1569          *
1570          * @return string
1571          */
1572         private static function replaceUnusualEscapesCallback( $matches ) {
1573                 $char = urldecode( $matches[0] );
1574                 $ord = ord( $char );
1575                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1576                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1577                         # No, shouldn't be escaped
1578                         return $char;
1579                 } else {
1580                         # Yes, leave it escaped
1581                         return $matches[0];
1582                 }
1583         }
1584
1585         /**
1586          * make an image if it's allowed, either through the global
1587          * option, through the exception, or through the on-wiki whitelist
1588          * @private
1589          *
1590          * $param $url string
1591          *
1592          * @return string
1593          */
1594         function maybeMakeExternalImage( $url ) {
1595                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1596                 $imagesexception = !empty( $imagesfrom );
1597                 $text = false;
1598                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1599                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1600                         $imagematch = false;
1601                         foreach ( $imagesfrom as $match ) {
1602                                 if ( strpos( $url, $match ) === 0 ) {
1603                                         $imagematch = true;
1604                                         break;
1605                                 }
1606                         }
1607                 } elseif ( $imagesexception ) {
1608                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1609                 } else {
1610                         $imagematch = false;
1611                 }
1612                 if ( $this->mOptions->getAllowExternalImages()
1613                          || ( $imagesexception && $imagematch ) ) {
1614                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1615                                 # Image found
1616                                 $text = Linker::makeExternalImage( $url );
1617                         }
1618                 }
1619                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1620                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1621                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1622                         foreach ( $whitelist as $entry ) {
1623                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1624                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1625                                         continue;
1626                                 }
1627                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1628                                         # Image matches a whitelist entry
1629                                         $text = Linker::makeExternalImage( $url );
1630                                         break;
1631                                 }
1632                         }
1633                 }
1634                 return $text;
1635         }
1636
1637         /**
1638          * Process [[ ]] wikilinks
1639          *
1640          * @param $s string
1641          *
1642          * @return String: processed text
1643          *
1644          * @private
1645          */
1646         function replaceInternalLinks( $s ) {
1647                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1648                 return $s;
1649         }
1650
1651         /**
1652          * Process [[ ]] wikilinks (RIL)
1653          * @return LinkHolderArray
1654          *
1655          * @private
1656          */
1657         function replaceInternalLinks2( &$s ) {
1658                 wfProfileIn( __METHOD__ );
1659
1660                 wfProfileIn( __METHOD__.'-setup' );
1661                 static $tc = FALSE, $e1, $e1_img;
1662                 # the % is needed to support urlencoded titles as well
1663                 if ( !$tc ) {
1664                         $tc = Title::legalChars() . '#%';
1665                         # Match a link having the form [[namespace:link|alternate]]trail
1666                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1667                         # Match cases where there is no "]]", which might still be images
1668                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1669                 }
1670
1671                 $holders = new LinkHolderArray( $this );
1672
1673                 # split the entire text string on occurences of [[
1674                 $a = StringUtils::explode( '[[', ' ' . $s );
1675                 # get the first element (all text up to first [[), and remove the space we added
1676                 $s = $a->current();
1677                 $a->next();
1678                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1679                 $s = substr( $s, 1 );
1680
1681                 $useLinkPrefixExtension = $this->getFunctionLang()->linkPrefixExtension();
1682                 $e2 = null;
1683                 if ( $useLinkPrefixExtension ) {
1684                         # Match the end of a line for a word that's not followed by whitespace,
1685                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1686                         $e2 = wfMsgForContent( 'linkprefix' );
1687                 }
1688
1689                 if ( is_null( $this->mTitle ) ) {
1690                         wfProfileOut( __METHOD__.'-setup' );
1691                         wfProfileOut( __METHOD__ );
1692                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1693                 }
1694                 $nottalk = !$this->mTitle->isTalkPage();
1695
1696                 if ( $useLinkPrefixExtension ) {
1697                         $m = array();
1698                         if ( preg_match( $e2, $s, $m ) ) {
1699                                 $first_prefix = $m[2];
1700                         } else {
1701                                 $first_prefix = false;
1702                         }
1703                 } else {
1704                         $prefix = '';
1705                 }
1706
1707                 if ( $this->getFunctionLang()->hasVariants() ) {
1708                         $selflink = $this->getFunctionLang()->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1709                 } else {
1710                         $selflink = array( $this->mTitle->getPrefixedText() );
1711                 }
1712                 $useSubpages = $this->areSubpagesAllowed();
1713                 wfProfileOut( __METHOD__.'-setup' );
1714
1715                 # Loop for each link
1716                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1717                         # Check for excessive memory usage
1718                         if ( $holders->isBig() ) {
1719                                 # Too big
1720                                 # Do the existence check, replace the link holders and clear the array
1721                                 $holders->replace( $s );
1722                                 $holders->clear();
1723                         }
1724
1725                         if ( $useLinkPrefixExtension ) {
1726                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1727                                 if ( preg_match( $e2, $s, $m ) ) {
1728                                         $prefix = $m[2];
1729                                         $s = $m[1];
1730                                 } else {
1731                                         $prefix='';
1732                                 }
1733                                 # first link
1734                                 if ( $first_prefix ) {
1735                                         $prefix = $first_prefix;
1736                                         $first_prefix = false;
1737                                 }
1738                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1739                         }
1740
1741                         $might_be_img = false;
1742
1743                         wfProfileIn( __METHOD__."-e1" );
1744                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1745                                 $text = $m[2];
1746                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1747                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1748                                 # the real problem is with the $e1 regex
1749                                 # See bug 1300.
1750                                 #
1751                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1752                                 # and no image is in sight. See bug 2095.
1753                                 #
1754                                 if ( $text !== '' &&
1755                                         substr( $m[3], 0, 1 ) === ']' &&
1756                                         strpos( $text, '[' ) !== false
1757                                 )
1758                                 {
1759                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1760                                         $m[3] = substr( $m[3], 1 );
1761                                 }
1762                                 # fix up urlencoded title texts
1763                                 if ( strpos( $m[1], '%' ) !== false ) {
1764                                         # Should anchors '#' also be rejected?
1765                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1766                                 }
1767                                 $trail = $m[3];
1768                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1769                                 $might_be_img = true;
1770                                 $text = $m[2];
1771                                 if ( strpos( $m[1], '%' ) !== false ) {
1772                                         $m[1] = rawurldecode( $m[1] );
1773                                 }
1774                                 $trail = "";
1775                         } else { # Invalid form; output directly
1776                                 $s .= $prefix . '[[' . $line ;
1777                                 wfProfileOut( __METHOD__."-e1" );
1778                                 continue;
1779                         }
1780                         wfProfileOut( __METHOD__."-e1" );
1781                         wfProfileIn( __METHOD__."-misc" );
1782
1783                         # Don't allow internal links to pages containing
1784                         # PROTO: where PROTO is a valid URL protocol; these
1785                         # should be external links.
1786                         if ( preg_match( '/^(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1787                                 $s .= $prefix . '[[' . $line ;
1788                                 wfProfileOut( __METHOD__."-misc" );
1789                                 continue;
1790                         }
1791
1792                         # Make subpage if necessary
1793                         if ( $useSubpages ) {
1794                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1795                         } else {
1796                                 $link = $m[1];
1797                         }
1798
1799                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1800                         if ( !$noforce ) {
1801                                 # Strip off leading ':'
1802                                 $link = substr( $link, 1 );
1803                         }
1804
1805                         wfProfileOut( __METHOD__."-misc" );
1806                         wfProfileIn( __METHOD__."-title" );
1807                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1808                         if ( $nt === null ) {
1809                                 $s .= $prefix . '[[' . $line;
1810                                 wfProfileOut( __METHOD__."-title" );
1811                                 continue;
1812                         }
1813
1814                         $ns = $nt->getNamespace();
1815                         $iw = $nt->getInterWiki();
1816                         wfProfileOut( __METHOD__."-title" );
1817
1818                         if ( $might_be_img ) { # if this is actually an invalid link
1819                                 wfProfileIn( __METHOD__."-might_be_img" );
1820                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1821                                         $found = false;
1822                                         while ( true ) {
1823                                                 # look at the next 'line' to see if we can close it there
1824                                                 $a->next();
1825                                                 $next_line = $a->current();
1826                                                 if ( $next_line === false || $next_line === null ) {
1827                                                         break;
1828                                                 }
1829                                                 $m = explode( ']]', $next_line, 3 );
1830                                                 if ( count( $m ) == 3 ) {
1831                                                         # the first ]] closes the inner link, the second the image
1832                                                         $found = true;
1833                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1834                                                         $trail = $m[2];
1835                                                         break;
1836                                                 } elseif ( count( $m ) == 2 ) {
1837                                                         # if there's exactly one ]] that's fine, we'll keep looking
1838                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1839                                                 } else {
1840                                                         # if $next_line is invalid too, we need look no further
1841                                                         $text .= '[[' . $next_line;
1842                                                         break;
1843                                                 }
1844                                         }
1845                                         if ( !$found ) {
1846                                                 # we couldn't find the end of this imageLink, so output it raw
1847                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1848                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1849                                                 $s .= "{$prefix}[[$link|$text";
1850                                                 # note: no $trail, because without an end, there *is* no trail
1851                                                 wfProfileOut( __METHOD__."-might_be_img" );
1852                                                 continue;
1853                                         }
1854                                 } else { # it's not an image, so output it raw
1855                                         $s .= "{$prefix}[[$link|$text";
1856                                         # note: no $trail, because without an end, there *is* no trail
1857                                         wfProfileOut( __METHOD__."-might_be_img" );
1858                                         continue;
1859                                 }
1860                                 wfProfileOut( __METHOD__."-might_be_img" );
1861                         }
1862
1863                         $wasblank = ( $text  == '' );
1864                         if ( $wasblank ) {
1865                                 $text = $link;
1866                         } else {
1867                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1868                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1869                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1870                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1871                                 $text = $this->doQuotes( $text );
1872                         }
1873
1874                         # Link not escaped by : , create the various objects
1875                         if ( $noforce ) {
1876                                 global $wgContLang;
1877
1878                                 # Interwikis
1879                                 wfProfileIn( __METHOD__."-interwiki" );
1880                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1881                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1882                                         $s = rtrim( $s . $prefix );
1883                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1884                                         wfProfileOut( __METHOD__."-interwiki" );
1885                                         continue;
1886                                 }
1887                                 wfProfileOut( __METHOD__."-interwiki" );
1888
1889                                 if ( $ns == NS_FILE ) {
1890                                         wfProfileIn( __METHOD__."-image" );
1891                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1892                                                 if ( $wasblank ) {
1893                                                         # if no parameters were passed, $text
1894                                                         # becomes something like "File:Foo.png",
1895                                                         # which we don't want to pass on to the
1896                                                         # image generator
1897                                                         $text = '';
1898                                                 } else {
1899                                                         # recursively parse links inside the image caption
1900                                                         # actually, this will parse them in any other parameters, too,
1901                                                         # but it might be hard to fix that, and it doesn't matter ATM
1902                                                         $text = $this->replaceExternalLinks( $text );
1903                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1904                                                 }
1905                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1906                                                 $s .= $prefix . $this->armorLinks(
1907                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1908                                         } else {
1909                                                 $s .= $prefix . $trail;
1910                                         }
1911                                         wfProfileOut( __METHOD__."-image" );
1912                                         continue;
1913                                 }
1914
1915                                 if ( $ns == NS_CATEGORY ) {
1916                                         wfProfileIn( __METHOD__."-category" );
1917                                         $s = rtrim( $s . "\n" ); # bug 87
1918
1919                                         if ( $wasblank ) {
1920                                                 $sortkey = $this->getDefaultSort();
1921                                         } else {
1922                                                 $sortkey = $text;
1923                                         }
1924                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1925                                         $sortkey = str_replace( "\n", '', $sortkey );
1926                                         $sortkey = $this->getFunctionLang()->convertCategoryKey( $sortkey );
1927                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1928
1929                                         /**
1930                                          * Strip the whitespace Category links produce, see bug 87
1931                                          * @todo We might want to use trim($tmp, "\n") here.
1932                                          */
1933                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1934
1935                                         wfProfileOut( __METHOD__."-category" );
1936                                         continue;
1937                                 }
1938                         }
1939
1940                         # Self-link checking
1941                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1942                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1943                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
1944                                         continue;
1945                                 }
1946                         }
1947
1948                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1949                         # @todo FIXME: Should do batch file existence checks, see comment below
1950                         if ( $ns == NS_MEDIA ) {
1951                                 wfProfileIn( __METHOD__."-media" );
1952                                 # Give extensions a chance to select the file revision for us
1953                                 $options = array();
1954                                 $descQuery = false;
1955                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
1956                                         array( $this, $nt, &$options, &$descQuery ) );
1957                                 # Fetch and register the file (file title may be different via hooks)
1958                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
1959                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1960                                 $s .= $prefix . $this->armorLinks(
1961                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
1962                                 wfProfileOut( __METHOD__."-media" );
1963                                 continue;
1964                         }
1965
1966                         wfProfileIn( __METHOD__."-always_known" );
1967                         # Some titles, such as valid special pages or files in foreign repos, should
1968                         # be shown as bluelinks even though they're not included in the page table
1969                         #
1970                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1971                         # batch file existence checks for NS_FILE and NS_MEDIA
1972                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1973                                 $this->mOutput->addLink( $nt );
1974                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
1975                         } else {
1976                                 # Links will be added to the output link list after checking
1977                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
1978                         }
1979                         wfProfileOut( __METHOD__."-always_known" );
1980                 }
1981                 wfProfileOut( __METHOD__ );
1982                 return $holders;
1983         }
1984
1985         /**
1986          * Render a forced-blue link inline; protect against double expansion of
1987          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1988          * Since this little disaster has to split off the trail text to avoid
1989          * breaking URLs in the following text without breaking trails on the
1990          * wiki links, it's been made into a horrible function.
1991          *
1992          * @param $nt Title
1993          * @param $text String
1994          * @param $query Array or String
1995          * @param $trail String
1996          * @param $prefix String
1997          * @return String: HTML-wikitext mix oh yuck
1998          */
1999         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2000                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2001
2002                 if ( is_string( $query ) ) {
2003                         $query = wfCgiToArray( $query );
2004                 }
2005                 if ( $text == '' ) {
2006                         $text = htmlspecialchars( $nt->getPrefixedText() );
2007                 }
2008
2009                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2010
2011                 return $this->armorLinks( $link ) . $trail;
2012         }
2013
2014         /**
2015          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2016          * going to go through further parsing steps before inline URL expansion.
2017          *
2018          * Not needed quite as much as it used to be since free links are a bit
2019          * more sensible these days. But bracketed links are still an issue.
2020          *
2021          * @param $text String: more-or-less HTML
2022          * @return String: less-or-more HTML with NOPARSE bits
2023          */
2024         function armorLinks( $text ) {
2025                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2026                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2027         }
2028
2029         /**
2030          * Return true if subpage links should be expanded on this page.
2031          * @return Boolean
2032          */
2033         function areSubpagesAllowed() {
2034                 # Some namespaces don't allow subpages
2035                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2036         }
2037
2038         /**
2039          * Handle link to subpage if necessary
2040          *
2041          * @param $target String: the source of the link
2042          * @param &$text String: the link text, modified as necessary
2043          * @return string the full name of the link
2044          * @private
2045          */
2046         function maybeDoSubpageLink( $target, &$text ) {
2047                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2048         }
2049
2050         /**#@+
2051          * Used by doBlockLevels()
2052          * @private
2053          *
2054          * @return string
2055          */
2056         function closeParagraph() {
2057                 $result = '';
2058                 if ( $this->mLastSection != '' ) {
2059                         $result = '</' . $this->mLastSection  . ">\n";
2060                 }
2061                 $this->mInPre = false;
2062                 $this->mLastSection = '';
2063                 return $result;
2064         }
2065
2066         /**
2067          * getCommon() returns the length of the longest common substring
2068          * of both arguments, starting at the beginning of both.
2069          * @private
2070          *
2071          * @param $st1 string
2072          * @param $st2 string
2073          *
2074          * @return int
2075          */
2076         function getCommon( $st1, $st2 ) {
2077                 $fl = strlen( $st1 );
2078                 $shorter = strlen( $st2 );
2079                 if ( $fl < $shorter ) {
2080                         $shorter = $fl;
2081                 }
2082
2083                 for ( $i = 0; $i < $shorter; ++$i ) {
2084                         if ( $st1[$i] != $st2[$i] ) {
2085                                 break;
2086                         }
2087                 }
2088                 return $i;
2089         }
2090
2091         /**
2092          * These next three functions open, continue, and close the list
2093          * element appropriate to the prefix character passed into them.
2094          * @private
2095          *
2096          * @param $char char
2097          *
2098          * @return string
2099          */
2100         function openList( $char ) {
2101                 $result = $this->closeParagraph();
2102
2103                 if ( '*' === $char ) {
2104                         $result .= '<ul><li>';
2105                 } elseif ( '#' === $char ) {
2106                         $result .= '<ol><li>';
2107                 } elseif ( ':' === $char ) {
2108                         $result .= '<dl><dd>';
2109                 } elseif ( ';' === $char ) {
2110                         $result .= '<dl><dt>';
2111                         $this->mDTopen = true;
2112                 } else {
2113                         $result = '<!-- ERR 1 -->';
2114                 }
2115
2116                 return $result;
2117         }
2118
2119         /**
2120          * TODO: document
2121          * @param $char String
2122          * @private
2123          *
2124          * @return string
2125          */
2126         function nextItem( $char ) {
2127                 if ( '*' === $char || '#' === $char ) {
2128                         return '</li><li>';
2129                 } elseif ( ':' === $char || ';' === $char ) {
2130                         $close = '</dd>';
2131                         if ( $this->mDTopen ) {
2132                                 $close = '</dt>';
2133                         }
2134                         if ( ';' === $char ) {
2135                                 $this->mDTopen = true;
2136                                 return $close . '<dt>';
2137                         } else {
2138                                 $this->mDTopen = false;
2139                                 return $close . '<dd>';
2140                         }
2141                 }
2142                 return '<!-- ERR 2 -->';
2143         }
2144
2145         /**
2146          * TODO: document
2147          * @param $char String
2148          * @private
2149          *
2150          * @return string
2151          */
2152         function closeList( $char ) {
2153                 if ( '*' === $char ) {
2154                         $text = '</li></ul>';
2155                 } elseif ( '#' === $char ) {
2156                         $text = '</li></ol>';
2157                 } elseif ( ':' === $char ) {
2158                         if ( $this->mDTopen ) {
2159                                 $this->mDTopen = false;
2160                                 $text = '</dt></dl>';
2161                         } else {
2162                                 $text = '</dd></dl>';
2163                         }
2164                 } else {
2165                         return '<!-- ERR 3 -->';
2166                 }
2167                 return $text."\n";
2168         }
2169         /**#@-*/
2170
2171         /**
2172          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2173          *
2174          * @param $text String
2175          * @param $linestart Boolean: whether or not this is at the start of a line.
2176          * @private
2177          * @return string the lists rendered as HTML
2178          */
2179         function doBlockLevels( $text, $linestart ) {
2180                 wfProfileIn( __METHOD__ );
2181
2182                 # Parsing through the text line by line.  The main thing
2183                 # happening here is handling of block-level elements p, pre,
2184                 # and making lists from lines starting with * # : etc.
2185                 #
2186                 $textLines = StringUtils::explode( "\n", $text );
2187
2188                 $lastPrefix = $output = '';
2189                 $this->mDTopen = $inBlockElem = false;
2190                 $prefixLength = 0;
2191                 $paragraphStack = false;
2192
2193                 foreach ( $textLines as $oLine ) {
2194                         # Fix up $linestart
2195                         if ( !$linestart ) {
2196                                 $output .= $oLine;
2197                                 $linestart = true;
2198                                 continue;
2199                         }
2200                         # * = ul
2201                         # # = ol
2202                         # ; = dt
2203                         # : = dd
2204
2205                         $lastPrefixLength = strlen( $lastPrefix );
2206                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2207                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2208                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2209                         if ( !$this->mInPre ) {
2210                                 # Multiple prefixes may abut each other for nested lists.
2211                                 $prefixLength = strspn( $oLine, '*#:;' );
2212                                 $prefix = substr( $oLine, 0, $prefixLength );
2213
2214                                 # eh?
2215                                 # ; and : are both from definition-lists, so they're equivalent
2216                                 #  for the purposes of determining whether or not we need to open/close
2217                                 #  elements.
2218                                 $prefix2 = str_replace( ';', ':', $prefix );
2219                                 $t = substr( $oLine, $prefixLength );
2220                                 $this->mInPre = (bool)$preOpenMatch;
2221                         } else {
2222                                 # Don't interpret any other prefixes in preformatted text
2223                                 $prefixLength = 0;
2224                                 $prefix = $prefix2 = '';
2225                                 $t = $oLine;
2226                         }
2227
2228                         # List generation
2229                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2230                                 # Same as the last item, so no need to deal with nesting or opening stuff
2231                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2232                                 $paragraphStack = false;
2233
2234                                 if ( substr( $prefix, -1 ) === ';') {
2235                                         # The one nasty exception: definition lists work like this:
2236                                         # ; title : definition text
2237                                         # So we check for : in the remainder text to split up the
2238                                         # title and definition, without b0rking links.
2239                                         $term = $t2 = '';
2240                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2241                                                 $t = $t2;
2242                                                 $output .= $term . $this->nextItem( ':' );
2243                                         }
2244                                 }
2245                         } elseif ( $prefixLength || $lastPrefixLength ) {
2246                                 # We need to open or close prefixes, or both.
2247
2248                                 # Either open or close a level...
2249                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2250                                 $paragraphStack = false;
2251
2252                                 # Close all the prefixes which aren't shared.
2253                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2254                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2255                                         --$lastPrefixLength;
2256                                 }
2257
2258                                 # Continue the current prefix if appropriate.
2259                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2260                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2261                                 }
2262
2263                                 # Open prefixes where appropriate.
2264                                 while ( $prefixLength > $commonPrefixLength ) {
2265                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2266                                         $output .= $this->openList( $char );
2267
2268                                         if ( ';' === $char ) {
2269                                                 # @todo FIXME: This is dupe of code above
2270                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2271                                                         $t = $t2;
2272                                                         $output .= $term . $this->nextItem( ':' );
2273                                                 }
2274                                         }
2275                                         ++$commonPrefixLength;
2276                                 }
2277                                 $lastPrefix = $prefix2;
2278                         }
2279
2280                         # If we have no prefixes, go to paragraph mode.
2281                         if ( 0 == $prefixLength ) {
2282                                 wfProfileIn( __METHOD__."-paragraph" );
2283                                 # No prefix (not in list)--go to paragraph mode
2284                                 # XXX: use a stack for nestable elements like span, table and div
2285                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2286                                 $closematch = preg_match(
2287                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2288                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2289                                 if ( $openmatch or $closematch ) {
2290                                         $paragraphStack = false;
2291                                         # TODO bug 5718: paragraph closed
2292                                         $output .= $this->closeParagraph();
2293                                         if ( $preOpenMatch and !$preCloseMatch ) {
2294                                                 $this->mInPre = true;
2295                                         }
2296                                         $inBlockElem = !$closematch;
2297                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2298                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2299                                                 # pre
2300                                                 if ( $this->mLastSection !== 'pre' ) {
2301                                                         $paragraphStack = false;
2302                                                         $output .= $this->closeParagraph().'<pre>';
2303                                                         $this->mLastSection = 'pre';
2304                                                 }
2305                                                 $t = substr( $t, 1 );
2306                                         } else {
2307                                                 # paragraph
2308                                                 if ( trim( $t ) === '' ) {
2309                                                         if ( $paragraphStack ) {
2310                                                                 $output .= $paragraphStack.'<br />';
2311                                                                 $paragraphStack = false;
2312                                                                 $this->mLastSection = 'p';
2313                                                         } else {
2314                                                                 if ( $this->mLastSection !== 'p' ) {
2315                                                                         $output .= $this->closeParagraph();
2316                                                                         $this->mLastSection = '';
2317                                                                         $paragraphStack = '<p>';
2318                                                                 } else {
2319                                                                         $paragraphStack = '</p><p>';
2320                                                                 }
2321                                                         }
2322                                                 } else {
2323                                                         if ( $paragraphStack ) {
2324                                                                 $output .= $paragraphStack;
2325                                                                 $paragraphStack = false;
2326                                                                 $this->mLastSection = 'p';
2327                                                         } elseif ( $this->mLastSection !== 'p' ) {
2328                                                                 $output .= $this->closeParagraph().'<p>';
2329                                                                 $this->mLastSection = 'p';
2330                                                         }
2331                                                 }
2332                                         }
2333                                 }
2334                                 wfProfileOut( __METHOD__."-paragraph" );
2335                         }
2336                         # somewhere above we forget to get out of pre block (bug 785)
2337                         if ( $preCloseMatch && $this->mInPre ) {
2338                                 $this->mInPre = false;
2339                         }
2340                         if ( $paragraphStack === false ) {
2341                                 $output .= $t."\n";
2342                         }
2343                 }
2344                 while ( $prefixLength ) {
2345                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2346                         --$prefixLength;
2347                 }
2348                 if ( $this->mLastSection != '' ) {
2349                         $output .= '</' . $this->mLastSection . '>';
2350                         $this->mLastSection = '';
2351                 }
2352
2353                 wfProfileOut( __METHOD__ );
2354                 return $output;
2355         }
2356
2357         /**
2358          * Split up a string on ':', ignoring any occurences inside tags
2359          * to prevent illegal overlapping.
2360          *
2361          * @param $str String the string to split
2362          * @param &$before String set to everything before the ':'
2363          * @param &$after String set to everything after the ':'
2364          * @return String the position of the ':', or false if none found
2365          */
2366         function findColonNoLinks( $str, &$before, &$after ) {
2367                 wfProfileIn( __METHOD__ );
2368
2369                 $pos = strpos( $str, ':' );
2370                 if ( $pos === false ) {
2371                         # Nothing to find!
2372                         wfProfileOut( __METHOD__ );
2373                         return false;
2374                 }
2375
2376                 $lt = strpos( $str, '<' );
2377                 if ( $lt === false || $lt > $pos ) {
2378                         # Easy; no tag nesting to worry about
2379                         $before = substr( $str, 0, $pos );
2380                         $after = substr( $str, $pos+1 );
2381                         wfProfileOut( __METHOD__ );
2382                         return $pos;
2383                 }
2384
2385                 # Ugly state machine to walk through avoiding tags.
2386                 $state = self::COLON_STATE_TEXT;
2387                 $stack = 0;
2388                 $len = strlen( $str );
2389                 for( $i = 0; $i < $len; $i++ ) {
2390                         $c = $str[$i];
2391
2392                         switch( $state ) {
2393                         # (Using the number is a performance hack for common cases)
2394                         case 0: # self::COLON_STATE_TEXT:
2395                                 switch( $c ) {
2396                                 case "<":
2397                                         # Could be either a <start> tag or an </end> tag
2398                                         $state = self::COLON_STATE_TAGSTART;
2399                                         break;
2400                                 case ":":
2401                                         if ( $stack == 0 ) {
2402                                                 # We found it!
2403                                                 $before = substr( $str, 0, $i );
2404                                                 $after = substr( $str, $i + 1 );
2405                                                 wfProfileOut( __METHOD__ );
2406                                                 return $i;
2407                                         }
2408                                         # Embedded in a tag; don't break it.
2409                                         break;
2410                                 default:
2411                                         # Skip ahead looking for something interesting
2412                                         $colon = strpos( $str, ':', $i );
2413                                         if ( $colon === false ) {
2414                                                 # Nothing else interesting
2415                                                 wfProfileOut( __METHOD__ );
2416                                                 return false;
2417                                         }
2418                                         $lt = strpos( $str, '<', $i );
2419                                         if ( $stack === 0 ) {
2420                                                 if ( $lt === false || $colon < $lt ) {
2421                                                         # We found it!
2422                                                         $before = substr( $str, 0, $colon );
2423                                                         $after = substr( $str, $colon + 1 );
2424                                                         wfProfileOut( __METHOD__ );
2425                                                         return $i;
2426                                                 }
2427                                         }
2428                                         if ( $lt === false ) {
2429                                                 # Nothing else interesting to find; abort!
2430                                                 # We're nested, but there's no close tags left. Abort!
2431                                                 break 2;
2432                                         }
2433                                         # Skip ahead to next tag start
2434                                         $i = $lt;
2435                                         $state = self::COLON_STATE_TAGSTART;
2436                                 }
2437                                 break;
2438                         case 1: # self::COLON_STATE_TAG:
2439                                 # In a <tag>
2440                                 switch( $c ) {
2441                                 case ">":
2442                                         $stack++;
2443                                         $state = self::COLON_STATE_TEXT;
2444                                         break;
2445                                 case "/":
2446                                         # Slash may be followed by >?
2447                                         $state = self::COLON_STATE_TAGSLASH;
2448                                         break;
2449                                 default:
2450                                         # ignore
2451                                 }
2452                                 break;
2453                         case 2: # self::COLON_STATE_TAGSTART:
2454                                 switch( $c ) {
2455                                 case "/":
2456                                         $state = self::COLON_STATE_CLOSETAG;
2457                                         break;
2458                                 case "!":
2459                                         $state = self::COLON_STATE_COMMENT;
2460                                         break;
2461                                 case ">":
2462                                         # Illegal early close? This shouldn't happen D:
2463                                         $state = self::COLON_STATE_TEXT;
2464                                         break;
2465                                 default:
2466                                         $state = self::COLON_STATE_TAG;
2467                                 }
2468                                 break;
2469                         case 3: # self::COLON_STATE_CLOSETAG:
2470                                 # In a </tag>
2471                                 if ( $c === ">" ) {
2472                                         $stack--;
2473                                         if ( $stack < 0 ) {
2474                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2475                                                 wfProfileOut( __METHOD__ );
2476                                                 return false;
2477                                         }
2478                                         $state = self::COLON_STATE_TEXT;
2479                                 }
2480                                 break;
2481                         case self::COLON_STATE_TAGSLASH:
2482                                 if ( $c === ">" ) {
2483                                         # Yes, a self-closed tag <blah/>
2484                                         $state = self::COLON_STATE_TEXT;
2485                                 } else {
2486                                         # Probably we're jumping the gun, and this is an attribute
2487                                         $state = self::COLON_STATE_TAG;
2488                                 }
2489                                 break;
2490                         case 5: # self::COLON_STATE_COMMENT:
2491                                 if ( $c === "-" ) {
2492                                         $state = self::COLON_STATE_COMMENTDASH;
2493                                 }
2494                                 break;
2495                         case self::COLON_STATE_COMMENTDASH:
2496                                 if ( $c === "-" ) {
2497                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2498                                 } else {
2499                                         $state = self::COLON_STATE_COMMENT;
2500                                 }
2501                                 break;
2502                         case self::COLON_STATE_COMMENTDASHDASH:
2503                                 if ( $c === ">" ) {
2504                                         $state = self::COLON_STATE_TEXT;
2505                                 } else {
2506                                         $state = self::COLON_STATE_COMMENT;
2507                                 }
2508                                 break;
2509                         default:
2510                                 throw new MWException( "State machine error in " . __METHOD__ );
2511                         }
2512                 }
2513                 if ( $stack > 0 ) {
2514                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2515                         wfProfileOut( __METHOD__ );
2516                         return false;
2517                 }
2518                 wfProfileOut( __METHOD__ );
2519                 return false;
2520         }
2521
2522         /**
2523          * Return value of a magic variable (like PAGENAME)
2524          *
2525          * @private
2526          *
2527          * @param $index integer
2528          * @param $frame PPFrame
2529          *
2530          * @return string
2531          */
2532         function getVariableValue( $index, $frame = false ) {
2533                 global $wgContLang, $wgSitename, $wgServer;
2534                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2535
2536                 if ( is_null( $this->mTitle ) ) {
2537                         // If no title set, bad things are going to happen
2538                         // later. Title should always be set since this
2539                         // should only be called in the middle of a parse
2540                         // operation (but the unit-tests do funky stuff)
2541                         throw new MWException( __METHOD__ . ' Should only be '
2542                                 . ' called while parsing (no title set)' );
2543                 }
2544
2545                 /**
2546                  * Some of these require message or data lookups and can be
2547                  * expensive to check many times.
2548                  */
2549                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2550                         if ( isset( $this->mVarCache[$index] ) ) {
2551                                 return $this->mVarCache[$index];
2552                         }
2553                 }
2554
2555                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2556                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2557
2558                 # Use the time zone
2559                 global $wgLocaltimezone;
2560                 if ( isset( $wgLocaltimezone ) ) {
2561                         $oldtz = date_default_timezone_get();
2562                         date_default_timezone_set( $wgLocaltimezone );
2563                 }
2564
2565                 $localTimestamp = date( 'YmdHis', $ts );
2566                 $localMonth = date( 'm', $ts );
2567                 $localMonth1 = date( 'n', $ts );
2568                 $localMonthName = date( 'n', $ts );
2569                 $localDay = date( 'j', $ts );
2570                 $localDay2 = date( 'd', $ts );
2571                 $localDayOfWeek = date( 'w', $ts );
2572                 $localWeek = date( 'W', $ts );
2573                 $localYear = date( 'Y', $ts );
2574                 $localHour = date( 'H', $ts );
2575                 if ( isset( $wgLocaltimezone ) ) {
2576                         date_default_timezone_set( $oldtz );
2577                 }
2578
2579                 $pageLang = $this->getFunctionLang();
2580
2581                 switch ( $index ) {
2582                         case 'currentmonth':
2583                                 $value = $pageLang->formatNum( gmdate( 'm', $ts ) );
2584                                 break;
2585                         case 'currentmonth1':
2586                                 $value = $pageLang->formatNum( gmdate( 'n', $ts ) );
2587                                 break;
2588                         case 'currentmonthname':
2589                                 $value = $pageLang->getMonthName( gmdate( 'n', $ts ) );
2590                                 break;
2591                         case 'currentmonthnamegen':
2592                                 $value = $pageLang->getMonthNameGen( gmdate( 'n', $ts ) );
2593                                 break;
2594                         case 'currentmonthabbrev':
2595                                 $value = $pageLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2596                                 break;
2597                         case 'currentday':
2598                                 $value = $pageLang->formatNum( gmdate( 'j', $ts ) );
2599                                 break;
2600                         case 'currentday2':
2601                                 $value = $pageLang->formatNum( gmdate( 'd', $ts ) );
2602                                 break;
2603                         case 'localmonth':
2604                                 $value = $pageLang->formatNum( $localMonth );
2605                                 break;
2606                         case 'localmonth1':
2607                                 $value = $pageLang->formatNum( $localMonth1 );
2608                                 break;
2609                         case 'localmonthname':
2610                                 $value = $pageLang->getMonthName( $localMonthName );
2611                                 break;
2612                         case 'localmonthnamegen':
2613                                 $value = $pageLang->getMonthNameGen( $localMonthName );
2614                                 break;
2615                         case 'localmonthabbrev':
2616                                 $value = $pageLang->getMonthAbbreviation( $localMonthName );
2617                                 break;
2618                         case 'localday':
2619                                 $value = $pageLang->formatNum( $localDay );
2620                                 break;
2621                         case 'localday2':
2622                                 $value = $pageLang->formatNum( $localDay2 );
2623                                 break;
2624                         case 'pagename':
2625                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2626                                 break;
2627                         case 'pagenamee':
2628                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2629                                 break;
2630                         case 'fullpagename':
2631                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2632                                 break;
2633                         case 'fullpagenamee':
2634                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2635                                 break;
2636                         case 'subpagename':
2637                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2638                                 break;
2639                         case 'subpagenamee':
2640                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2641                                 break;
2642                         case 'basepagename':
2643                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2644                                 break;
2645                         case 'basepagenamee':
2646                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2647                                 break;
2648                         case 'talkpagename':
2649                                 if ( $this->mTitle->canTalk() ) {
2650                                         $talkPage = $this->mTitle->getTalkPage();
2651                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2652                                 } else {
2653                                         $value = '';
2654                                 }
2655                                 break;
2656                         case 'talkpagenamee':
2657                                 if ( $this->mTitle->canTalk() ) {
2658                                         $talkPage = $this->mTitle->getTalkPage();
2659                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2660                                 } else {
2661                                         $value = '';
2662                                 }
2663                                 break;
2664                         case 'subjectpagename':
2665                                 $subjPage = $this->mTitle->getSubjectPage();
2666                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2667                                 break;
2668                         case 'subjectpagenamee':
2669                                 $subjPage = $this->mTitle->getSubjectPage();
2670                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2671                                 break;
2672                         case 'revisionid':
2673                                 # Let the edit saving system know we should parse the page
2674                                 # *after* a revision ID has been assigned.
2675                                 $this->mOutput->setFlag( 'vary-revision' );
2676                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2677                                 $value = $this->mRevisionId;
2678                                 break;
2679                         case 'revisionday':
2680                                 # Let the edit saving system know we should parse the page
2681                                 # *after* a revision ID has been assigned. This is for null edits.
2682                                 $this->mOutput->setFlag( 'vary-revision' );
2683                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2684                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2685                                 break;
2686                         case 'revisionday2':
2687                                 # Let the edit saving system know we should parse the page
2688                                 # *after* a revision ID has been assigned. This is for null edits.
2689                                 $this->mOutput->setFlag( 'vary-revision' );
2690                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2691                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2692                                 break;
2693                         case 'revisionmonth':
2694                                 # Let the edit saving system know we should parse the page
2695                                 # *after* a revision ID has been assigned. This is for null edits.
2696                                 $this->mOutput->setFlag( 'vary-revision' );
2697                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2698                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2699                                 break;
2700                         case 'revisionmonth1':
2701                                 # Let the edit saving system know we should parse the page
2702                                 # *after* a revision ID has been assigned. This is for null edits.
2703                                 $this->mOutput->setFlag( 'vary-revision' );
2704                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2705                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2706                                 break;
2707                         case 'revisionyear':
2708                                 # Let the edit saving system know we should parse the page
2709                                 # *after* a revision ID has been assigned. This is for null edits.
2710                                 $this->mOutput->setFlag( 'vary-revision' );
2711                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2712                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2713                                 break;
2714                         case 'revisiontimestamp':
2715                                 # Let the edit saving system know we should parse the page
2716                                 # *after* a revision ID has been assigned. This is for null edits.
2717                                 $this->mOutput->setFlag( 'vary-revision' );
2718                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2719                                 $value = $this->getRevisionTimestamp();
2720                                 break;
2721                         case 'revisionuser':
2722                                 # Let the edit saving system know we should parse the page
2723                                 # *after* a revision ID has been assigned. This is for null edits.
2724                                 $this->mOutput->setFlag( 'vary-revision' );
2725                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2726                                 $value = $this->getRevisionUser();
2727                                 break;
2728                         case 'namespace':
2729                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2730                                 break;
2731                         case 'namespacee':
2732                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2733                                 break;
2734                         case 'talkspace':
2735                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2736                                 break;
2737                         case 'talkspacee':
2738                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2739                                 break;
2740                         case 'subjectspace':
2741                                 $value = $this->mTitle->getSubjectNsText();
2742                                 break;
2743                         case 'subjectspacee':
2744                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2745                                 break;
2746                         case 'currentdayname':
2747                                 $value = $pageLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2748                                 break;
2749                         case 'currentyear':
2750                                 $value = $pageLang->formatNum( gmdate( 'Y', $ts ), true );
2751                                 break;
2752                         case 'currenttime':
2753                                 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2754                                 break;
2755                         case 'currenthour':
2756                                 $value = $pageLang->formatNum( gmdate( 'H', $ts ), true );
2757                                 break;
2758                         case 'currentweek':
2759                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2760                                 # int to remove the padding
2761                                 $value = $pageLang->formatNum( (int)gmdate( 'W', $ts ) );
2762                                 break;
2763                         case 'currentdow':
2764                                 $value = $pageLang->formatNum( gmdate( 'w', $ts ) );
2765                                 break;
2766                         case 'localdayname':
2767                                 $value = $pageLang->getWeekdayName( $localDayOfWeek + 1 );
2768                                 break;
2769                         case 'localyear':
2770                                 $value = $pageLang->formatNum( $localYear, true );
2771                                 break;
2772                         case 'localtime':
2773                                 $value = $pageLang->time( $localTimestamp, false, false );
2774                                 break;
2775                         case 'localhour':
2776                                 $value = $pageLang->formatNum( $localHour, true );
2777                                 break;
2778                         case 'localweek':
2779                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2780                                 # int to remove the padding
2781                                 $value = $pageLang->formatNum( (int)$localWeek );
2782                                 break;
2783                         case 'localdow':
2784                                 $value = $pageLang->formatNum( $localDayOfWeek );
2785                                 break;
2786                         case 'numberofarticles':
2787                                 $value = $pageLang->formatNum( SiteStats::articles() );
2788                                 break;
2789                         case 'numberoffiles':
2790                                 $value = $pageLang->formatNum( SiteStats::images() );
2791                                 break;
2792                         case 'numberofusers':
2793                                 $value = $pageLang->formatNum( SiteStats::users() );
2794                                 break;
2795                         case 'numberofactiveusers':
2796                                 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2797                                 break;
2798                         case 'numberofpages':
2799                                 $value = $pageLang->formatNum( SiteStats::pages() );
2800                                 break;
2801                         case 'numberofadmins':
2802                                 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2803                                 break;
2804                         case 'numberofedits':
2805                                 $value = $pageLang->formatNum( SiteStats::edits() );
2806                                 break;
2807                         case 'numberofviews':
2808                                 $value = $pageLang->formatNum( SiteStats::views() );
2809                                 break;
2810                         case 'currenttimestamp':
2811                                 $value = wfTimestamp( TS_MW, $ts );
2812                                 break;
2813                         case 'localtimestamp':
2814                                 $value = $localTimestamp;
2815                                 break;
2816                         case 'currentversion':
2817                                 $value = SpecialVersion::getVersion();
2818                                 break;
2819                         case 'articlepath':
2820                                 return $wgArticlePath;
2821                         case 'sitename':
2822                                 return $wgSitename;
2823                         case 'server':
2824                                 return $wgServer;
2825                         case 'servername':
2826                                 $serverParts = wfParseUrl( $wgServer );
2827                                 return $serverParts && isset( $serverParts['host'] ) ? $serverParts['host'] : $wgServer;
2828                         case 'scriptpath':
2829                                 return $wgScriptPath;
2830                         case 'stylepath':
2831                                 return $wgStylePath;
2832                         case 'directionmark':
2833                                 return $pageLang->getDirMark();
2834                         case 'contentlanguage':
2835                                 global $wgLanguageCode;
2836                                 return $wgLanguageCode;
2837                         default:
2838                                 $ret = null;
2839                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2840                                         return $ret;
2841                                 } else {
2842                                         return null;
2843                                 }
2844                 }
2845
2846                 if ( $index ) {
2847                         $this->mVarCache[$index] = $value;
2848                 }
2849
2850                 return $value;
2851         }
2852
2853         /**
2854          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2855          *
2856          * @private
2857          */
2858         function initialiseVariables() {
2859                 wfProfileIn( __METHOD__ );
2860                 $variableIDs = MagicWord::getVariableIDs();
2861                 $substIDs = MagicWord::getSubstIDs();
2862
2863                 $this->mVariables = new MagicWordArray( $variableIDs );
2864                 $this->mSubstWords = new MagicWordArray( $substIDs );
2865                 wfProfileOut( __METHOD__ );
2866         }
2867
2868         /**
2869          * Preprocess some wikitext and return the document tree.
2870          * This is the ghost of replace_variables().
2871          *
2872          * @param $text String: The text to parse
2873          * @param $flags Integer: bitwise combination of:
2874          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2875          *                                     included. Default is to assume a direct page view.
2876          *
2877          * The generated DOM tree must depend only on the input text and the flags.
2878          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2879          *
2880          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2881          * change in the DOM tree for a given text, must be passed through the section identifier
2882          * in the section edit link and thus back to extractSections().
2883          *
2884          * The output of this function is currently only cached in process memory, but a persistent
2885          * cache may be implemented at a later date which takes further advantage of these strict
2886          * dependency requirements.
2887          *
2888          * @private
2889          *
2890          * @return PPNode
2891          */
2892         function preprocessToDom( $text, $flags = 0 ) {
2893                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2894                 return $dom;
2895         }
2896
2897         /**
2898          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2899          *
2900          * @param $s string
2901          *
2902          * @return array
2903          */
2904         public static function splitWhitespace( $s ) {
2905                 $ltrimmed = ltrim( $s );
2906                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2907                 $trimmed = rtrim( $ltrimmed );
2908                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2909                 if ( $diff > 0 ) {
2910                         $w2 = substr( $ltrimmed, -$diff );
2911                 } else {
2912                         $w2 = '';
2913                 }
2914                 return array( $w1, $trimmed, $w2 );
2915         }
2916
2917         /**
2918          * Replace magic variables, templates, and template arguments
2919          * with the appropriate text. Templates are substituted recursively,
2920          * taking care to avoid infinite loops.
2921          *
2922          * Note that the substitution depends on value of $mOutputType:
2923          *  self::OT_WIKI: only {{subst:}} templates
2924          *  self::OT_PREPROCESS: templates but not extension tags
2925          *  self::OT_HTML: all templates and extension tags
2926          *
2927          * @param $text String the text to transform
2928          * @param $frame PPFrame Object describing the arguments passed to the template.
2929          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2930          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2931          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
2932          * @private
2933          *
2934          * @return string
2935          */
2936         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2937                 # Is there any text? Also, Prevent too big inclusions!
2938                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2939                         return $text;
2940                 }
2941                 wfProfileIn( __METHOD__ );
2942
2943                 if ( $frame === false ) {
2944                         $frame = $this->getPreprocessor()->newFrame();
2945                 } elseif ( !( $frame instanceof PPFrame ) ) {
2946                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2947                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2948                 }
2949
2950                 $dom = $this->preprocessToDom( $text );
2951                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2952                 $text = $frame->expand( $dom, $flags );
2953
2954                 wfProfileOut( __METHOD__ );
2955                 return $text;
2956         }
2957
2958         /**
2959          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2960          *
2961          * @param $args array
2962          *
2963          * @return array
2964          */
2965         static function createAssocArgs( $args ) {
2966                 $assocArgs = array();
2967                 $index = 1;
2968                 foreach ( $args as $arg ) {
2969                         $eqpos = strpos( $arg, '=' );
2970                         if ( $eqpos === false ) {
2971                                 $assocArgs[$index++] = $arg;
2972                         } else {
2973                                 $name = trim( substr( $arg, 0, $eqpos ) );
2974                                 $value = trim( substr( $arg, $eqpos+1 ) );
2975                                 if ( $value === false ) {
2976                                         $value = '';
2977                                 }
2978                                 if ( $name !== false ) {
2979                                         $assocArgs[$name] = $value;
2980                                 }
2981                         }
2982                 }
2983
2984                 return $assocArgs;
2985         }
2986
2987         /**
2988          * Warn the user when a parser limitation is reached
2989          * Will warn at most once the user per limitation type
2990          *
2991          * @param $limitationType String: should be one of:
2992          *   'expensive-parserfunction' (corresponding messages:
2993          *       'expensive-parserfunction-warning',
2994          *       'expensive-parserfunction-category')
2995          *   'post-expand-template-argument' (corresponding messages:
2996          *       'post-expand-template-argument-warning',
2997          *       'post-expand-template-argument-category')
2998          *   'post-expand-template-inclusion' (corresponding messages:
2999          *       'post-expand-template-inclusion-warning',
3000          *       'post-expand-template-inclusion-category')
3001          * @param $current Current value
3002          * @param $max Maximum allowed, when an explicit limit has been
3003          *       exceeded, provide the values (optional)
3004          */
3005         function limitationWarn( $limitationType, $current=null, $max=null) {
3006                 # does no harm if $current and $max are present but are unnecessary for the message
3007                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3008                 $this->mOutput->addWarning( $warning );
3009                 $this->addTrackingCategory( "$limitationType-category" );
3010         }
3011
3012         /**
3013          * Return the text of a template, after recursively
3014          * replacing any variables or templates within the template.
3015          *
3016          * @param $piece Array: the parts of the template
3017          *  $piece['title']: the title, i.e. the part before the |
3018          *  $piece['parts']: the parameter array
3019          *  $piece['lineStart']: whether the brace was at the start of a line
3020          * @param $frame PPFrame The current frame, contains template arguments
3021          * @return String: the text of the template
3022          * @private
3023          */
3024         function braceSubstitution( $piece, $frame ) {
3025                 global $wgNonincludableNamespaces, $wgEnableInterwikiTranscluding, $wgEnableInterwikiTemplatesTracking;
3026                 wfProfileIn( __METHOD__ );
3027                 wfProfileIn( __METHOD__.'-setup' );
3028
3029                 # Flags
3030                 $found = false;             # $text has been filled
3031                 $nowiki = false;            # wiki markup in $text should be escaped
3032                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3033                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3034                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3035
3036                 # Title object, where $text came from
3037                 $title = null;
3038
3039                 # $part1 is the bit before the first |, and must contain only title characters.
3040                 # Various prefixes will be stripped from it later.
3041                 $titleWithSpaces = $frame->expand( $piece['title'] );
3042                 $part1 = trim( $titleWithSpaces );
3043                 $titleText = false;
3044
3045                 # Original title text preserved for various purposes
3046                 $originalTitle = $part1;
3047
3048                 # $args is a list of argument nodes, starting from index 0, not including $part1
3049                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3050                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3051                 wfProfileOut( __METHOD__.'-setup' );
3052                 wfProfileIn( __METHOD__."-title-$originalTitle" );
3053
3054                 # SUBST
3055                 wfProfileIn( __METHOD__.'-modifiers' );
3056                 if ( !$found ) {
3057
3058                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3059
3060                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3061                         # Decide whether to expand template or keep wikitext as-is.
3062                         if ( $this->ot['wiki'] ) {
3063                                 if ( $substMatch === false ) {
3064                                         $literal = true;  # literal when in PST with no prefix
3065                                 } else {
3066                                         $literal = false; # expand when in PST with subst: or safesubst:
3067                                 }
3068                         } else {
3069                                 if ( $substMatch == 'subst' ) {
3070                                         $literal = true;  # literal when not in PST with plain subst:
3071                                 } else {
3072                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3073                                 }
3074                         }
3075                         if ( $literal ) {
3076                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3077                                 $isLocalObj = true;
3078                                 $found = true;
3079                         }
3080                 }
3081
3082                 # Variables
3083                 if ( !$found && $args->getLength() == 0 ) {
3084                         $id = $this->mVariables->matchStartToEnd( $part1 );
3085                         if ( $id !== false ) {
3086                                 $text = $this->getVariableValue( $id, $frame );
3087                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3088                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3089                                 }
3090                                 $found = true;
3091                         }
3092                 }
3093
3094                 # MSG, MSGNW and RAW
3095                 if ( !$found ) {
3096                         # Check for MSGNW:
3097                         $mwMsgnw = MagicWord::get( 'msgnw' );
3098                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3099                                 $nowiki = true;
3100                         } else {
3101                                 # Remove obsolete MSG:
3102                                 $mwMsg = MagicWord::get( 'msg' );
3103                                 $mwMsg->matchStartAndRemove( $part1 );
3104                         }
3105
3106                         # Check for RAW:
3107                         $mwRaw = MagicWord::get( 'raw' );
3108                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3109                                 $forceRawInterwiki = true;
3110                         }
3111                 }
3112                 wfProfileOut( __METHOD__.'-modifiers' );
3113
3114                 # Parser functions
3115                 if ( !$found ) {
3116                         wfProfileIn( __METHOD__ . '-pfunc' );
3117
3118                         $colonPos = strpos( $part1, ':' );
3119                         if ( $colonPos !== false ) {
3120                                 # Case sensitive functions
3121                                 $function = substr( $part1, 0, $colonPos );
3122                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3123                                         $function = $this->mFunctionSynonyms[1][$function];
3124                                 } else {
3125                                         # Case insensitive functions
3126                                         $function = $this->getFunctionLang()->lc( $function );
3127                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3128                                                 $function = $this->mFunctionSynonyms[0][$function];
3129                                         } else {
3130                                                 $function = false;
3131                                         }
3132                                 }
3133                                 if ( $function ) {
3134                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3135                                         $initialArgs = array( &$this );
3136                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3137                                         if ( $flags & SFH_OBJECT_ARGS ) {
3138                                                 # Add a frame parameter, and pass the arguments as an array
3139                                                 $allArgs = $initialArgs;
3140                                                 $allArgs[] = $frame;
3141                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3142                                                         $funcArgs[] = $args->item( $i );
3143                                                 }
3144                                                 $allArgs[] = $funcArgs;
3145                                         } else {
3146                                                 # Convert arguments to plain text
3147                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3148                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3149                                                 }
3150                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3151                                         }
3152
3153                                         # Workaround for PHP bug 35229 and similar
3154                                         if ( !is_callable( $callback ) ) {
3155                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3156                                                 wfProfileOut( __METHOD__ );
3157                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3158                                         }
3159                                         $result = call_user_func_array( $callback, $allArgs );
3160                                         $found = true;
3161                                         $noparse = true;
3162                                         $preprocessFlags = 0;
3163
3164                                         if ( is_array( $result ) ) {
3165                                                 if ( isset( $result[0] ) ) {
3166                                                         $text = $result[0];
3167                                                         unset( $result[0] );
3168                                                 }
3169
3170                                                 # Extract flags into the local scope
3171                                                 # This allows callers to set flags such as nowiki, found, etc.
3172                                                 extract( $result );
3173                                         } else {
3174                                                 $text = $result;
3175                                         }
3176                                         if ( !$noparse ) {
3177                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3178                                                 $isChildObj = true;
3179                                         }
3180                                 }
3181                         }
3182                         wfProfileOut( __METHOD__ . '-pfunc' );
3183                 }
3184
3185                 # Finish mangling title and then check for loops.
3186                 # Set $title to a Title object and $titleText to the PDBK
3187                 if ( !$found ) {
3188                         $ns = NS_TEMPLATE;
3189                         # Split the title into page and subpage
3190                         $subpage = '';
3191                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3192                         if ( $subpage !== '' ) {
3193                                 $ns = $this->mTitle->getNamespace();
3194                         }
3195                         $title = Title::newFromText( $part1, $ns );
3196                         if ( $title ) {
3197                                 if ( !$title->isExternal() && $piece['interwiki'] !== '' ) {
3198                                         $title->setInterwiki( $piece['interwiki'] );
3199                                 }
3200                                 $titleText = $title->getPrefixedText();
3201                                 # Check for language variants if the template is not found
3202                                 if ( $this->getFunctionLang()->hasVariants() && $title->getArticleID() == 0 ) {
3203                                         $this->getFunctionLang()->findVariantLink( $part1, $title, true );
3204                                 }
3205                                 # Do recursion depth check
3206                                 $limit = $this->mOptions->getMaxTemplateDepth();
3207                                 if ( $frame->depth >= $limit ) {
3208                                         $found = true;
3209                                         $text = '<span class="error">'
3210                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3211                                                 . '</span>';
3212                                 }
3213                         }
3214                 }
3215
3216                 # Load from database
3217                 if ( !$found && $title ) {
3218                         wfProfileIn( __METHOD__ . '-loadtpl' );
3219                         if ( !$title->isExternal() ) {
3220                                 if ( $title->getNamespace() == NS_SPECIAL
3221                                         && $this->mOptions->getAllowSpecialInclusion()
3222                                         && $this->ot['html'] )
3223                                 {
3224                                         // Pass the template arguments as URL parameters.
3225                                         // "uselang" will have no effect since the Language object
3226                                         // is forced to the one defined in ParserOptions.
3227                                         $pageArgs = array();
3228                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3229                                                 $bits = $args->item( $i )->splitArg();
3230                                                 if ( strval( $bits['index'] ) === '' ) {
3231                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3232                                                         $value = trim( $frame->expand( $bits['value'] ) );
3233                                                         $pageArgs[$name] = $value;
3234                                                 }
3235                                         }
3236
3237                                         // Create a new context to execute the special page
3238                                         $context = new RequestContext;
3239                                         $context->setTitle( $title );
3240                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3241                                         $context->setUser( $this->getUser() );
3242                                         $context->setLang( Language::factory( $this->mOptions->getUserLang() ) );
3243                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3244                                         if ( $ret ) {
3245                                                 $text = $context->getOutput()->getHTML();
3246                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3247                                                 $found = true;
3248                                                 $isHTML = true;
3249                                                 $this->disableCache();
3250                                         }
3251                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3252                                         $found = false; # access denied
3253                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3254                                 } else {
3255                                         list( $text, $title ) = $this->getTemplateDom( $title );
3256                                         if ( $text !== false ) {
3257                                                 $found = true;
3258                                                 $isChildObj = true;
3259                                         }
3260                                 }
3261
3262                                 # If the title is valid but undisplayable, make a link to it
3263                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3264                                         $text = "[[:$titleText]]";
3265                                         $found = true;
3266                                 }
3267                         } elseif ( $wgEnableInterwikiTranscluding && $title->isTrans() ) {
3268
3269                                 $text = Interwiki::interwikiTransclude( $title );
3270                                 $this->registerDistantTemplate( $title );
3271
3272                                 if ( $wgEnableInterwikiTemplatesTracking ) {
3273                                         $this->registerDistantTemplate( $title );
3274                                 }
3275
3276                                 if ( $text !== false ) {
3277                                         # Preprocess it like a template
3278                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3279                                         $found = true;
3280                                         $isChildObj = true;
3281                                 }
3282
3283                         }
3284
3285                         # Do infinite loop check
3286                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3287                         if ( !$frame->loopCheck( $title ) ) {
3288                                 $found = true;
3289                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3290                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3291                         }
3292                         wfProfileOut( __METHOD__ . '-loadtpl' );
3293                 }
3294
3295                 # If we haven't found text to substitute by now, we're done
3296                 # Recover the source wikitext and return it
3297                 if ( !$found ) {
3298                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3299                         wfProfileOut( __METHOD__."-title-$originalTitle" );
3300                         wfProfileOut( __METHOD__ );
3301                         return array( 'object' => $text );
3302                 }
3303
3304                 # Expand DOM-style return values in a child frame
3305                 if ( $isChildObj ) {
3306                         # Clean up argument array
3307                         $newFrame = $frame->newChild( $args, $title );
3308
3309                         if ( $nowiki ) {
3310                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3311                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3312                                 # Expansion is eligible for the empty-frame cache
3313                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3314                                         $text = $this->mTplExpandCache[$titleText];
3315                                 } else {
3316                                         $text = $newFrame->expand( $text );
3317                                         $this->mTplExpandCache[$titleText] = $text;
3318                                 }
3319                         } else {
3320                                 # Uncached expansion
3321                                 $text = $newFrame->expand( $text );
3322                         }
3323                 }
3324                 if ( $isLocalObj && $nowiki ) {
3325                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3326                         $isLocalObj = false;
3327                 }
3328
3329                 # Replace raw HTML by a placeholder
3330                 # Add a blank line preceding, to prevent it from mucking up
3331                 # immediately preceding headings
3332                 if ( $isHTML ) {
3333                         $text = "\n\n" . $this->insertStripItem( $text );
3334                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3335                         # Escape nowiki-style return values
3336                         $text = wfEscapeWikiText( $text );
3337                 } elseif ( is_string( $text )
3338                         && !$piece['lineStart']
3339                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3340                 {
3341                         # Bug 529: if the template begins with a table or block-level
3342                         # element, it should be treated as beginning a new line.
3343                         # This behaviour is somewhat controversial.
3344                         $text = "\n" . $text;
3345                 }
3346
3347                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3348                         # Error, oversize inclusion
3349                         if ( $titleText !== false ) {
3350                                 # Make a working, properly escaped link if possible (bug 23588)
3351                                 $text = "[[:$titleText]]";
3352                         } else {
3353                                 # This will probably not be a working link, but at least it may
3354                                 # provide some hint of where the problem is
3355                                 preg_replace( '/^:/', '', $originalTitle );
3356                                 $text = "[[:$originalTitle]]";
3357                         }
3358                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3359                         $this->limitationWarn( 'post-expand-template-inclusion' );
3360                 }
3361
3362                 if ( $isLocalObj ) {
3363                         $ret = array( 'object' => $text );
3364                 } else {
3365                         $ret = array( 'text' => $text );
3366                 }
3367
3368                 wfProfileOut( __METHOD__."-title-$originalTitle" );
3369                 wfProfileOut( __METHOD__ );
3370                 return $ret;
3371         }
3372
3373         /**
3374          * Get the semi-parsed DOM representation of a template with a given title,
3375          * and its redirect destination title. Cached.
3376          *
3377          * @param $title Title
3378          *
3379          * @return array
3380          */
3381         function getTemplateDom( $title ) {
3382                 $cacheTitle = $title;
3383                 $titleText = $title->getPrefixedDBkey();
3384
3385                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3386                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3387                         $title = Title::makeTitle( $ns, $dbk );
3388                         $titleText = $title->getPrefixedDBkey();
3389                 }
3390                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3391                         return array( $this->mTplDomCache[$titleText], $title );
3392                 }
3393
3394                 # Cache miss, go to the database
3395                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3396
3397                 if ( $text === false ) {
3398                         $this->mTplDomCache[$titleText] = false;
3399                         return array( false, $title );
3400                 }
3401
3402                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3403                 $this->mTplDomCache[ $titleText ] = $dom;
3404
3405                 if ( !$title->equals( $cacheTitle ) ) {
3406                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3407                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3408                 }
3409
3410                 return array( $dom, $title );
3411         }
3412
3413         /**
3414          * Fetch the unparsed text of a template and register a reference to it.
3415          * @param Title $title
3416          * @return Array ( string or false, Title )
3417          */
3418         function fetchTemplateAndTitle( $title ) {
3419                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3420                 $stuff = call_user_func( $templateCb, $title, $this );
3421                 $text = $stuff['text'];
3422                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3423                 if ( isset( $stuff['deps'] ) ) {
3424                         foreach ( $stuff['deps'] as $dep ) {
3425                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3426                         }
3427                 }
3428                 return array( $text, $finalTitle );
3429         }
3430
3431         /**
3432          * Register a distant template as used
3433          */
3434         function registerDistantTemplate( $title ) {
3435                 $stuff = Parser::distantTemplateCallback( $title, $this );
3436                 $text = $stuff['text'];
3437                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3438                 if ( isset( $stuff['deps'] ) ) {
3439                         foreach ( $stuff['deps'] as $dep ) {
3440                                 $this->mOutput->addDistantTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3441                         }
3442                 }
3443         }
3444
3445         function fetchTemplate( $title ) {
3446                 $rv = $this->fetchTemplateAndTitle( $title );
3447                 return $rv[0];
3448         }
3449
3450         /**
3451          * Static function to get a template
3452          * Can be overridden via ParserOptions::setTemplateCallback().
3453          *
3454          * @parma $title Title
3455          * @param $parser Parser
3456          *
3457          * @return array
3458          */
3459         static function statelessFetchTemplate( $title, $parser = false ) {
3460                 $text = $skip = false;
3461                 $finalTitle = $title;
3462                 $deps = array();
3463
3464                 # Loop to fetch the article, with up to 1 redirect
3465                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3466                         # Give extensions a chance to select the revision instead
3467                         $id = false; # Assume current
3468                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3469                                 array( $parser, $title, &$skip, &$id ) );
3470
3471                         if ( $skip ) {
3472                                 $text = false;
3473                                 $deps[] = array(
3474                                         'title'         => $title,
3475                                         'page_id'       => $title->getArticleID(),
3476                                         'rev_id'        => null
3477                                 );
3478                                 break;
3479                         }
3480                         # Get the revision
3481                         $rev = $id
3482                                 ? Revision::newFromId( $id )
3483                                 : Revision::newFromTitle( $title );
3484                         $rev_id = $rev ? $rev->getId() : 0;
3485                         # If there is no current revision, there is no page
3486                         if ( $id === false && !$rev ) {
3487                                 $linkCache = LinkCache::singleton();
3488                                 $linkCache->addBadLinkObj( $title );
3489                         }
3490
3491                         $deps[] = array(
3492                                 'title'         => $title,
3493                                 'page_id'       => $title->getArticleID(),
3494                                 'rev_id'        => $rev_id );
3495                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3496                                 # We fetched a rev from a different title; register it too...
3497                                 $deps[] = array(
3498                                         'title'         => $rev->getTitle(),
3499                                         'page_id'       => $rev->getPage(),
3500                                         'rev_id'        => $rev_id );
3501                         }
3502
3503                         if ( $rev ) {
3504                                 $text = $rev->getText();
3505                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3506                                 global $wgContLang;
3507                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3508                                 if ( !$message->exists() ) {
3509                                         $text = false;
3510                                         break;
3511                                 }
3512                                 $text = $message->plain();
3513                         } else {
3514                                 break;
3515                         }
3516                         if ( $text === false ) {
3517                                 break;
3518                         }
3519                         # Redirect?
3520                         $finalTitle = $title;
3521                         $title = Title::newFromRedirect( $text );
3522                 }
3523                 return array(
3524                         'text' => $text,
3525                         'finalTitle' => $finalTitle,
3526                         'deps' => $deps );
3527         }
3528
3529         /**
3530          * Fetch a file and its title and register a reference to it.
3531          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3532          * @param Title $title
3533          * @param Array $options Array of options to RepoGroup::findFile
3534          * @return File|false
3535          */
3536         function fetchFile( $title, $options = array() ) {
3537                 $res = $this->fetchFileAndTitle( $title, $options );
3538                 return $res[0];
3539         }
3540
3541         /**
3542          * Fetch a file and its title and register a reference to it.
3543          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3544          * @param Title $title
3545          * @param Array $options Array of options to RepoGroup::findFile
3546          * @return Array ( File or false, Title of file )
3547          */
3548         function fetchFileAndTitle( $title, $options = array() ) {
3549                 if ( isset( $options['broken'] ) ) {
3550                         $file = false; // broken thumbnail forced by hook
3551                 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3552                         $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3553                 } else { // get by (name,timestamp)
3554                         $file = wfFindFile( $title, $options );
3555                 }
3556                 $time = $file ? $file->getTimestamp() : false;
3557                 $sha1 = $file ? $file->getSha1() : false;
3558                 # Register the file as a dependency...
3559                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3560                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3561                         # Update fetched file title
3562                         $title = $file->getTitle();
3563                         if ( is_null( $file->getRedirectedTitle() ) ) {
3564                                 # This file was not a redirect, but the title does not match.
3565                                 # Register under the new name because otherwise the link will
3566                                 # get lost.
3567                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3568                         }
3569                 }
3570                 return array( $file, $title );
3571         }
3572
3573         static function distantTemplateCallback( $title, $parser=false ) {
3574                 $text = '';
3575                 $rev_id = null;
3576                 $deps[] = array(
3577                         'title' => $title,
3578                         'page_id' => $title->getArticleID(),
3579                         'rev_id' => $rev_id );
3580
3581                 $finalTitle = $title;
3582
3583                 return array(
3584                         'text' => $text,
3585                         'finalTitle' => $finalTitle,
3586                         'deps' => $deps );
3587                 }
3588
3589         /**
3590          * Triple brace replacement -- used for template arguments
3591          * @private
3592          *
3593          * @param $peice array
3594          * @param $frame PPFrame
3595          *
3596          * @return array
3597          */
3598         function argSubstitution( $piece, $frame ) {
3599                 wfProfileIn( __METHOD__ );
3600
3601                 $error = false;
3602                 $parts = $piece['parts'];
3603                 $nameWithSpaces = $frame->expand( $piece['title'] );
3604                 $argName = trim( $nameWithSpaces );
3605                 $object = false;
3606                 $text = $frame->getArgument( $argName );
3607                 if (  $text === false && $parts->getLength() > 0
3608                   && (
3609                         $this->ot['html']
3610                         || $this->ot['pre']
3611                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3612                   )
3613                 ) {
3614                         # No match in frame, use the supplied default
3615                         $object = $parts->item( 0 )->getChildren();
3616                 }
3617                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3618                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3619                         $this->limitationWarn( 'post-expand-template-argument' );
3620                 }
3621
3622                 if ( $text === false && $object === false ) {
3623                         # No match anywhere
3624                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3625                 }
3626                 if ( $error !== false ) {
3627                         $text .= $error;
3628                 }
3629                 if ( $object !== false ) {
3630                         $ret = array( 'object' => $object );
3631                 } else {
3632                         $ret = array( 'text' => $text );
3633                 }
3634
3635                 wfProfileOut( __METHOD__ );
3636                 return $ret;
3637         }
3638
3639         /**
3640          * Return the text to be used for a given extension tag.
3641          * This is the ghost of strip().
3642          *
3643          * @param $params Associative array of parameters:
3644          *     name       PPNode for the tag name
3645          *     attr       PPNode for unparsed text where tag attributes are thought to be
3646          *     attributes Optional associative array of parsed attributes
3647          *     inner      Contents of extension element
3648          *     noClose    Original text did not have a close tag
3649          * @param $frame PPFrame
3650          *
3651          * @return string
3652          */
3653         function extensionSubstitution( $params, $frame ) {
3654                 $name = $frame->expand( $params['name'] );
3655                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3656                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3657                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3658
3659                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3660                         ( $this->ot['html'] || $this->ot['pre'] );
3661                 if ( $isFunctionTag ) {
3662                         $markerType = 'none';
3663                 } else {
3664                         $markerType = 'general';
3665                 }
3666                 if ( $this->ot['html'] || $isFunctionTag ) {
3667                         $name = strtolower( $name );
3668                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3669                         if ( isset( $params['attributes'] ) ) {
3670                                 $attributes = $attributes + $params['attributes'];
3671                         }
3672
3673                         if ( isset( $this->mTagHooks[$name] ) ) {
3674                                 # Workaround for PHP bug 35229 and similar
3675                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3676                                         throw new MWException( "Tag hook for $name is not callable\n" );
3677                                 }
3678                                 $output = call_user_func_array( $this->mTagHooks[$name],
3679                                         array( $content, $attributes, $this, $frame ) );
3680                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3681                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3682                                 if ( !is_callable( $callback ) ) {
3683                                         throw new MWException( "Tag hook for $name is not callable\n" );
3684                                 }
3685
3686                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3687                         } else {
3688                                 $output = '<span class="error">Invalid tag extension name: ' .
3689                                         htmlspecialchars( $name ) . '</span>';
3690                         }
3691
3692                         if ( is_array( $output ) ) {
3693                                 # Extract flags to local scope (to override $markerType)
3694                                 $flags = $output;
3695                                 $output = $flags[0];
3696                                 unset( $flags[0] );
3697                                 extract( $flags );
3698                         }
3699                 } else {
3700                         if ( is_null( $attrText ) ) {
3701                                 $attrText = '';
3702                         }
3703                         if ( isset( $params['attributes'] ) ) {
3704                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3705                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3706                                                 htmlspecialchars( $attrValue ) . '"';
3707                                 }
3708                         }
3709                         if ( $content === null ) {
3710                                 $output = "<$name$attrText/>";
3711                         } else {
3712                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3713                                 $output = "<$name$attrText>$content$close";
3714                         }
3715                 }
3716
3717                 if ( $markerType === 'none' ) {
3718                         return $output;
3719                 } elseif ( $markerType === 'nowiki' ) {
3720                         $this->mStripState->addNoWiki( $marker, $output );
3721                 } elseif ( $markerType === 'general' ) {
3722                         $this->mStripState->addGeneral( $marker, $output );
3723                 } else {
3724                         throw new MWException( __METHOD__.': invalid marker type' );
3725                 }
3726                 return $marker;
3727         }
3728
3729         /**
3730          * Increment an include size counter
3731          *
3732          * @param $type String: the type of expansion
3733          * @param $size Integer: the size of the text
3734          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3735          */
3736         function incrementIncludeSize( $type, $size ) {
3737                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3738                         return false;
3739                 } else {
3740                         $this->mIncludeSizes[$type] += $size;
3741                         return true;
3742                 }
3743         }
3744
3745         /**
3746          * Increment the expensive function count
3747          *
3748          * @return Boolean: false if the limit has been exceeded
3749          */
3750         function incrementExpensiveFunctionCount() {
3751                 global $wgExpensiveParserFunctionLimit;
3752                 $this->mExpensiveFunctionCount++;
3753                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3754                         return true;
3755                 }
3756                 return false;
3757         }
3758
3759         /**
3760          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3761          * Fills $this->mDoubleUnderscores, returns the modified text
3762          *
3763          * @param $text string
3764          *
3765          * @return string
3766          */
3767         function doDoubleUnderscore( $text ) {
3768                 wfProfileIn( __METHOD__ );
3769
3770                 # The position of __TOC__ needs to be recorded
3771                 $mw = MagicWord::get( 'toc' );
3772                 if ( $mw->match( $text ) ) {
3773                         $this->mShowToc = true;
3774                         $this->mForceTocPosition = true;
3775
3776                         # Set a placeholder. At the end we'll fill it in with the TOC.
3777                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3778
3779                         # Only keep the first one.
3780                         $text = $mw->replace( '', $text );
3781                 }
3782
3783                 # Now match and remove the rest of them
3784                 $mwa = MagicWord::getDoubleUnderscoreArray();
3785                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3786
3787                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3788                         $this->mOutput->mNoGallery = true;
3789                 }
3790                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3791                         $this->mShowToc = false;
3792                 }
3793                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3794                         $this->addTrackingCategory( 'hidden-category-category' );
3795                 }
3796                 # (bug 8068) Allow control over whether robots index a page.
3797                 #
3798                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3799                 # is not desirable, the last one on the page should win.
3800                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3801                         $this->mOutput->setIndexPolicy( 'noindex' );
3802                         $this->addTrackingCategory( 'noindex-category' );
3803                 }
3804                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3805                         $this->mOutput->setIndexPolicy( 'index' );
3806                         $this->addTrackingCategory( 'index-category' );
3807                 }
3808
3809                 # Cache all double underscores in the database
3810                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3811                         $this->mOutput->setProperty( $key, '' );
3812                 }
3813
3814                 wfProfileOut( __METHOD__ );
3815                 return $text;
3816         }
3817
3818         /**
3819          * Add a tracking category, getting the title from a system message,
3820          * or print a debug message if the title is invalid.
3821          *
3822          * @param $msg String: message key
3823          * @return Boolean: whether the addition was successful
3824          */
3825         protected function addTrackingCategory( $msg ) {
3826                 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
3827                         wfDebug( __METHOD__.": Not adding tracking category $msg to special page!\n" );
3828                         return false;
3829                 }
3830                 $cat = wfMsgForContent( $msg );
3831
3832                 # Allow tracking categories to be disabled by setting them to "-"
3833                 if ( $cat === '-' ) {
3834                         return false;
3835                 }
3836
3837                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3838                 if ( $containerCategory ) {
3839                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3840                         return true;
3841                 } else {
3842                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3843                         return false;
3844                 }
3845         }
3846
3847         /**
3848          * This function accomplishes several tasks:
3849          * 1) Auto-number headings if that option is enabled
3850          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3851          * 3) Add a Table of contents on the top for users who have enabled the option
3852          * 4) Auto-anchor headings
3853          *
3854          * It loops through all headlines, collects the necessary data, then splits up the
3855          * string and re-inserts the newly formatted headlines.
3856          *
3857          * @param $text String
3858          * @param $origText String: original, untouched wikitext
3859          * @param $isMain Boolean
3860          * @private
3861          */
3862         function formatHeadings( $text, $origText, $isMain=true ) {
3863                 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3864
3865                 # Inhibit editsection links if requested in the page
3866                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3867                         $showEditLink = 0;
3868                 } else {
3869                         $showEditLink = $this->mOptions->getEditSection();
3870                 }
3871                 if ( $showEditLink ) {
3872                         $this->mOutput->setEditSectionTokens( true );
3873                 }
3874
3875                 # Get all headlines for numbering them and adding funky stuff like [edit]
3876                 # links - this is for later, but we need the number of headlines right now
3877                 $matches = array();
3878                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3879
3880                 # if there are fewer than 4 headlines in the article, do not show TOC
3881                 # unless it's been explicitly enabled.
3882                 $enoughToc = $this->mShowToc &&
3883                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3884
3885                 # Allow user to stipulate that a page should have a "new section"
3886                 # link added via __NEWSECTIONLINK__
3887                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3888                         $this->mOutput->setNewSection( true );
3889                 }
3890
3891                 # Allow user to remove the "new section"
3892                 # link via __NONEWSECTIONLINK__
3893                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3894                         $this->mOutput->hideNewSection( true );
3895                 }
3896
3897                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3898                 # override above conditions and always show TOC above first header
3899                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3900                         $this->mShowToc = true;
3901                         $enoughToc = true;
3902                 }
3903
3904                 # headline counter
3905                 $headlineCount = 0;
3906                 $numVisible = 0;
3907
3908                 # Ugh .. the TOC should have neat indentation levels which can be
3909                 # passed to the skin functions. These are determined here
3910                 $toc = '';
3911                 $full = '';
3912                 $head = array();
3913                 $sublevelCount = array();
3914                 $levelCount = array();
3915                 $level = 0;
3916                 $prevlevel = 0;
3917                 $toclevel = 0;
3918                 $prevtoclevel = 0;
3919                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3920                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3921                 $oldType = $this->mOutputType;
3922                 $this->setOutputType( self::OT_WIKI );
3923                 $frame = $this->getPreprocessor()->newFrame();
3924                 $root = $this->preprocessToDom( $origText );
3925                 $node = $root->getFirstChild();
3926                 $byteOffset = 0;
3927                 $tocraw = array();
3928                 $refers = array();
3929
3930                 foreach ( $matches[3] as $headline ) {
3931                         $isTemplate = false;
3932                         $titleText = false;
3933                         $sectionIndex = false;
3934                         $numbering = '';
3935                         $markerMatches = array();
3936                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3937                                 $serial = $markerMatches[1];
3938                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3939                                 $isTemplate = ( $titleText != $baseTitleText );
3940                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3941                         }
3942
3943                         if ( $toclevel ) {
3944                                 $prevlevel = $level;
3945                         }
3946                         $level = $matches[1][$headlineCount];
3947
3948                         if ( $level > $prevlevel ) {
3949                                 # Increase TOC level
3950                                 $toclevel++;
3951                                 $sublevelCount[$toclevel] = 0;
3952                                 if ( $toclevel<$wgMaxTocLevel ) {
3953                                         $prevtoclevel = $toclevel;
3954                                         $toc .= Linker::tocIndent();
3955                                         $numVisible++;
3956                                 }
3957                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3958                                 # Decrease TOC level, find level to jump to
3959
3960                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3961                                         if ( $levelCount[$i] == $level ) {
3962                                                 # Found last matching level
3963                                                 $toclevel = $i;
3964                                                 break;
3965                                         } elseif ( $levelCount[$i] < $level ) {
3966                                                 # Found first matching level below current level
3967                                                 $toclevel = $i + 1;
3968                                                 break;
3969                                         }
3970                                 }
3971                                 if ( $i == 0 ) {
3972                                         $toclevel = 1;
3973                                 }
3974                                 if ( $toclevel<$wgMaxTocLevel ) {
3975                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3976                                                 # Unindent only if the previous toc level was shown :p
3977                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
3978                                                 $prevtoclevel = $toclevel;
3979                                         } else {
3980                                                 $toc .= Linker::tocLineEnd();
3981                                         }
3982                                 }
3983                         } else {
3984                                 # No change in level, end TOC line
3985                                 if ( $toclevel<$wgMaxTocLevel ) {
3986                                         $toc .= Linker::tocLineEnd();
3987                                 }
3988                         }
3989
3990                         $levelCount[$toclevel] = $level;
3991
3992                         # count number of headlines for each level
3993                         @$sublevelCount[$toclevel]++;
3994                         $dot = 0;
3995                         for( $i = 1; $i <= $toclevel; $i++ ) {
3996                                 if ( !empty( $sublevelCount[$i] ) ) {
3997                                         if ( $dot ) {
3998                                                 $numbering .= '.';
3999                                         }
4000                                         $numbering .= $this->getFunctionLang()->formatNum( $sublevelCount[$i] );
4001                                         $dot = 1;
4002                                 }
4003                         }
4004
4005                         # The safe header is a version of the header text safe to use for links
4006                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4007                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4008
4009                         # Remove link placeholders by the link text.
4010                         #     <!--LINK number-->
4011                         # turns into
4012                         #     link text with suffix
4013                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4014
4015                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4016                         $tocline = preg_replace(
4017                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4018                                 array( '',                          '<$1>' ),
4019                                 $safeHeadline
4020                         );
4021                         $tocline = trim( $tocline );
4022
4023                         # For the anchor, strip out HTML-y stuff period
4024                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4025                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4026
4027                         # Save headline for section edit hint before it's escaped
4028                         $headlineHint = $safeHeadline;
4029
4030                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4031                                 # For reverse compatibility, provide an id that's
4032                                 # HTML4-compatible, like we used to.
4033                                 #
4034                                 # It may be worth noting, academically, that it's possible for
4035                                 # the legacy anchor to conflict with a non-legacy headline
4036                                 # anchor on the page.  In this case likely the "correct" thing
4037                                 # would be to either drop the legacy anchors or make sure
4038                                 # they're numbered first.  However, this would require people
4039                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4040                                 # manually, so let's not bother worrying about it.
4041                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4042                                         array( 'noninitial', 'legacy' ) );
4043                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4044
4045                                 if ( $legacyHeadline == $safeHeadline ) {
4046                                         # No reason to have both (in fact, we can't)
4047                                         $legacyHeadline = false;
4048                                 }
4049                         } else {
4050                                 $legacyHeadline = false;
4051                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4052                                         'noninitial' );
4053                         }
4054
4055                         # HTML names must be case-insensitively unique (bug 10721).
4056                         # This does not apply to Unicode characters per
4057                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4058                         # @todo FIXME: We may be changing them depending on the current locale.
4059                         $arrayKey = strtolower( $safeHeadline );
4060                         if ( $legacyHeadline === false ) {
4061                                 $legacyArrayKey = false;
4062                         } else {
4063                                 $legacyArrayKey = strtolower( $legacyHeadline );
4064                         }
4065
4066                         # count how many in assoc. array so we can track dupes in anchors
4067                         if ( isset( $refers[$arrayKey] ) ) {
4068                                 $refers[$arrayKey]++;
4069                         } else {
4070                                 $refers[$arrayKey] = 1;
4071                         }
4072                         if ( isset( $refers[$legacyArrayKey] ) ) {
4073                                 $refers[$legacyArrayKey]++;
4074                         } else {
4075                                 $refers[$legacyArrayKey] = 1;
4076                         }
4077
4078                         # Don't number the heading if it is the only one (looks silly)
4079                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4080                                 # the two are different if the line contains a link
4081                                 $headline = $numbering . ' ' . $headline;
4082                         }
4083
4084                         # Create the anchor for linking from the TOC to the section
4085                         $anchor = $safeHeadline;
4086                         $legacyAnchor = $legacyHeadline;
4087                         if ( $refers[$arrayKey] > 1 ) {
4088                                 $anchor .= '_' . $refers[$arrayKey];
4089                         }
4090                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4091                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4092                         }
4093                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4094                                 $toc .= Linker::tocLine( $anchor, $tocline,
4095                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4096                         }
4097
4098                         # Add the section to the section tree
4099                         # Find the DOM node for this header
4100                         while ( $node && !$isTemplate ) {
4101                                 if ( $node->getName() === 'h' ) {
4102                                         $bits = $node->splitHeading();
4103                                         if ( $bits['i'] == $sectionIndex ) {
4104                                                 break;
4105                                         }
4106                                 }
4107                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4108                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4109                                 $node = $node->getNextSibling();
4110                         }
4111                         $tocraw[] = array(
4112                                 'toclevel' => $toclevel,
4113                                 'level' => $level,
4114                                 'line' => $tocline,
4115                                 'number' => $numbering,
4116                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4117                                 'fromtitle' => $titleText,
4118                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4119                                 'anchor' => $anchor,
4120                         );
4121
4122                         # give headline the correct <h#> tag
4123                         if ( $showEditLink && $sectionIndex !== false ) {
4124                                 // Output edit section links as markers with styles that can be customized by skins
4125                                 if ( $isTemplate ) {
4126                                         # Put a T flag in the section identifier, to indicate to extractSections()
4127                                         # that sections inside <includeonly> should be counted.
4128                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4129                                 } else {
4130                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4131                                 }
4132                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4133                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4134                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4135                                 // so we don't have to worry about a user trying to input one of these markers directly.
4136                                 // We use a page and section attribute to stop the language converter from converting these important bits
4137                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4138                                 // be able to convert that piece of data.
4139                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4140                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4141                                 if ( isset($editlinkArgs[2]) ) {
4142                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4143                                 } else {
4144                                         $editlink .= '/>';
4145                                 }
4146                         } else {
4147                                 $editlink = '';
4148                         }
4149                         $head[$headlineCount] = Linker::makeHeadline( $level,
4150                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4151                                 $editlink, $legacyAnchor );
4152
4153                         $headlineCount++;
4154                 }
4155
4156                 $this->setOutputType( $oldType );
4157
4158                 # Never ever show TOC if no headers
4159                 if ( $numVisible < 1 ) {
4160                         $enoughToc = false;
4161                 }
4162
4163                 if ( $enoughToc ) {
4164                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4165                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4166                         }
4167                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4168                         $this->mOutput->setTOCHTML( $toc );
4169                 }
4170
4171                 if ( $isMain ) {
4172                         $this->mOutput->setSections( $tocraw );
4173                 }
4174
4175                 # split up and insert constructed headlines
4176                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4177                 $i = 0;
4178
4179                 // build an array of document sections
4180                 $sections = array();
4181                 foreach ( $blocks as $block ) {
4182                         // $head is zero-based, sections aren't.
4183                         if ( empty( $head[$i - 1] ) ) {
4184                                 $sections[$i] = $block;
4185                         } else {
4186                                 $sections[$i] = $head[$i - 1] . $block;
4187                         }
4188
4189                         /**
4190                          * Send a hook, one per section.
4191                          * The idea here is to be able to make section-level DIVs, but to do so in a
4192                          * lower-impact, more correct way than r50769
4193                          *
4194                          * $this : caller
4195                          * $section : the section number
4196                          * &$sectionContent : ref to the content of the section
4197                          * $showEditLinks : boolean describing whether this section has an edit link
4198                          */
4199                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4200
4201                         $i++;
4202                 }
4203
4204                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4205                         // append the TOC at the beginning
4206                         // Top anchor now in skin
4207                         $sections[0] = $sections[0] . $toc . "\n";
4208                 }
4209
4210                 $full .= join( '', $sections );
4211
4212                 if ( $this->mForceTocPosition ) {
4213                         return str_replace( '<!--MWTOC-->', $toc, $full );
4214                 } else {
4215                         return $full;
4216                 }
4217         }
4218
4219         /**
4220          * Transform wiki markup when saving a page by doing \r\n -> \n
4221          * conversion, substitting signatures, {{subst:}} templates, etc.
4222          *
4223          * @param $text String: the text to transform
4224          * @param $title Title: the Title object for the current article
4225          * @param $user User: the User object describing the current user
4226          * @param $options ParserOptions: parsing options
4227          * @param $clearState Boolean: whether to clear the parser state first
4228          * @return String: the altered wiki markup
4229          */
4230         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4231                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4232                 $this->setUser( $user );
4233
4234                 $pairs = array(
4235                         "\r\n" => "\n",
4236                 );
4237                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4238                 if( $options->getPreSaveTransform() ) {
4239                         $text = $this->pstPass2( $text, $user );
4240                 }
4241                 $text = $this->mStripState->unstripBoth( $text );
4242
4243                 $this->setUser( null ); #Reset
4244
4245                 return $text;
4246         }
4247
4248         /**
4249          * Pre-save transform helper function
4250          * @private
4251          *
4252          * @param $text string
4253          * @param $user User
4254          *
4255          * @return string
4256          */
4257         function pstPass2( $text, $user ) {
4258                 global $wgContLang, $wgLocaltimezone;
4259
4260                 # Note: This is the timestamp saved as hardcoded wikitext to
4261                 # the database, we use $wgContLang here in order to give
4262                 # everyone the same signature and use the default one rather
4263                 # than the one selected in each user's preferences.
4264                 # (see also bug 12815)
4265                 $ts = $this->mOptions->getTimestamp();
4266                 if ( isset( $wgLocaltimezone ) ) {
4267                         $tz = $wgLocaltimezone;
4268                 } else {
4269                         $tz = date_default_timezone_get();
4270                 }
4271
4272                 $unixts = wfTimestamp( TS_UNIX, $ts );
4273                 $oldtz = date_default_timezone_get();
4274                 date_default_timezone_set( $tz );
4275                 $ts = date( 'YmdHis', $unixts );
4276                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4277
4278                 # Allow translation of timezones through wiki. date() can return
4279                 # whatever crap the system uses, localised or not, so we cannot
4280                 # ship premade translations.
4281                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4282                 $msg = wfMessage( $key )->inContentLanguage();
4283                 if ( $msg->exists() ) {
4284                         $tzMsg = $msg->text();
4285                 }
4286
4287                 date_default_timezone_set( $oldtz );
4288
4289                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4290
4291                 # Variable replacement
4292                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4293                 $text = $this->replaceVariables( $text );
4294
4295                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4296                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4297
4298                 # Signatures
4299                 $sigText = $this->getUserSig( $user );
4300                 $text = strtr( $text, array(
4301                         '~~~~~' => $d,
4302                         '~~~~' => "$sigText $d",
4303                         '~~~' => $sigText
4304                 ) );
4305
4306                 # Context links: [[|name]] and [[name (context)|]]
4307                 global $wgLegalTitleChars;
4308                 $tc = "[$wgLegalTitleChars]";
4309                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4310
4311                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";           # [[ns:page (context)|]]
4312                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";           # [[ns:page（context）|]]
4313                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4314                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4315
4316                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4317                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4318                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4319                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4320
4321                 $t = $this->mTitle->getText();
4322                 $m = array();
4323                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4324                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4325                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4326                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4327                 } else {
4328                         # if there's no context, don't bother duplicating the title
4329                         $text = preg_replace( $p2, '[[\\1]]', $text );
4330                 }
4331
4332                 # Trim trailing whitespace
4333                 $text = rtrim( $text );
4334
4335                 return $text;
4336         }
4337
4338         /**
4339          * Fetch the user's signature text, if any, and normalize to
4340          * validated, ready-to-insert wikitext.
4341          * If you have pre-fetched the nickname or the fancySig option, you can
4342          * specify them here to save a database query.
4343          * Do not reuse this parser instance after calling getUserSig(),
4344          * as it may have changed if it's the $wgParser.
4345          *
4346          * @param $user User
4347          * @param $nickname String|bool nickname to use or false to use user's default nickname
4348          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4349          *                  or null to use default value
4350          * @return string
4351          */
4352         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4353                 global $wgMaxSigChars;
4354
4355                 $username = $user->getName();
4356
4357                 # If not given, retrieve from the user object.
4358                 if ( $nickname === false )
4359                         $nickname = $user->getOption( 'nickname' );
4360
4361                 if ( is_null( $fancySig ) ) {
4362                         $fancySig = $user->getBoolOption( 'fancysig' );
4363                 }
4364
4365                 $nickname = $nickname == null ? $username : $nickname;
4366
4367                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4368                         $nickname = $username;
4369                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4370                 } elseif ( $fancySig !== false ) {
4371                         # Sig. might contain markup; validate this
4372                         if ( $this->validateSig( $nickname ) !== false ) {
4373                                 # Validated; clean up (if needed) and return it
4374                                 return $this->cleanSig( $nickname, true );
4375                         } else {
4376                                 # Failed to validate; fall back to the default
4377                                 $nickname = $username;
4378                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4379                         }
4380                 }
4381
4382                 # Make sure nickname doesnt get a sig in a sig
4383                 $nickname = $this->cleanSigInSig( $nickname );
4384
4385                 # If we're still here, make it a link to the user page
4386                 $userText = wfEscapeWikiText( $username );
4387                 $nickText = wfEscapeWikiText( $nickname );
4388                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4389
4390                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4391         }
4392
4393         /**
4394          * Check that the user's signature contains no bad XML
4395          *
4396          * @param $text String
4397          * @return mixed An expanded string, or false if invalid.
4398          */
4399         function validateSig( $text ) {
4400                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4401         }
4402
4403         /**
4404          * Clean up signature text
4405          *
4406          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4407          * 2) Substitute all transclusions
4408          *
4409          * @param $text String
4410          * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4411          * @return String: signature text
4412          */
4413         function cleanSig( $text, $parsing = false ) {
4414                 if ( !$parsing ) {
4415                         global $wgTitle;
4416                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4417                 }
4418
4419                 # Option to disable this feature
4420                 if ( !$this->mOptions->getCleanSignatures() ) {
4421                         return $text;
4422                 }
4423
4424                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4425                 #  => Move this logic to braceSubstitution()
4426                 $substWord = MagicWord::get( 'subst' );
4427                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4428                 $substText = '{{' . $substWord->getSynonym( 0 );
4429
4430                 $text = preg_replace( $substRegex, $substText, $text );
4431                 $text = $this->cleanSigInSig( $text );
4432                 $dom = $this->preprocessToDom( $text );
4433                 $frame = $this->getPreprocessor()->newFrame();
4434                 $text = $frame->expand( $dom );
4435
4436                 if ( !$parsing ) {
4437                         $text = $this->mStripState->unstripBoth( $text );
4438                 }
4439
4440                 return $text;
4441         }
4442
4443         /**
4444          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4445          *
4446          * @param $text String
4447          * @return String: signature text with /~{3,5}/ removed
4448          */
4449         function cleanSigInSig( $text ) {
4450                 $text = preg_replace( '/~{3,5}/', '', $text );
4451                 return $text;
4452         }
4453
4454         /**
4455          * Set up some variables which are usually set up in parse()
4456          * so that an external function can call some class members with confidence
4457          *
4458          * @param $title Title|null
4459          * @param $options ParserOptions
4460          * @param $outputType
4461          * @param $clearState bool
4462          */
4463         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4464                 $this->startParse( $title, $options, $outputType, $clearState );
4465         }
4466
4467         /**
4468          * @param $title Title|null
4469          * @param $options ParserOptions
4470          * @param $outputType
4471          * @param $clearState bool
4472          */
4473         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4474                 $this->setTitle( $title );
4475                 $this->mOptions = $options;
4476                 $this->setOutputType( $outputType );
4477                 if ( $clearState ) {
4478                         $this->clearState();
4479                 }
4480         }
4481
4482         /**
4483          * Wrapper for preprocess()
4484          *
4485          * @param $text String: the text to preprocess
4486          * @param $options ParserOptions: options
4487          * @param $title Title object or null to use $wgTitle
4488          * @return String
4489          */
4490         public function transformMsg( $text, $options, $title = null ) {
4491                 static $executing = false;
4492
4493                 # Guard against infinite recursion
4494                 if ( $executing ) {
4495                         return $text;
4496                 }
4497                 $executing = true;
4498
4499                 wfProfileIn( __METHOD__ );
4500                 if ( !$title ) {
4501                         global $wgTitle;
4502                         $title = $wgTitle;
4503                 }
4504                 if ( !$title ) {
4505                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4506                         # Create a ghost title in such case
4507                         $title = Title::newFromText( 'Dwimmerlaik' );
4508                 }
4509                 $text = $this->preprocess( $text, $title, $options );
4510
4511                 $executing = false;
4512                 wfProfileOut( __METHOD__ );
4513                 return $text;
4514         }
4515
4516         /**
4517          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4518          * The callback should have the following form:
4519          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4520          *
4521          * Transform and return $text. Use $parser for any required context, e.g. use
4522          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4523          *
4524          * Hooks may return extended information by returning an array, of which the
4525          * first numbered element (index 0) must be the return string, and all other
4526          * entries are extracted into local variables within an internal function
4527          * in the Parser class.
4528          *
4529          * This interface (introduced r61913) appears to be undocumented, but
4530          * 'markerName' is used by some core tag hooks to override which strip
4531          * array their results are placed in. **Use great caution if attempting
4532          * this interface, as it is not documented and injudicious use could smash
4533          * private variables.**
4534          *
4535          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4536          * @param $callback Mixed: the callback function (and object) to use for the tag
4537          * @return The old value of the mTagHooks array associated with the hook
4538          */
4539         public function setHook( $tag, $callback ) {
4540                 $tag = strtolower( $tag );
4541                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4542                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4543                 $this->mTagHooks[$tag] = $callback;
4544                 if ( !in_array( $tag, $this->mStripList ) ) {
4545                         $this->mStripList[] = $tag;
4546                 }
4547
4548                 return $oldVal;
4549         }
4550
4551         /**
4552          * As setHook(), but letting the contents be parsed.
4553          *
4554          * Transparent tag hooks are like regular XML-style tag hooks, except they
4555          * operate late in the transformation sequence, on HTML instead of wikitext.
4556          *
4557          * This is probably obsoleted by things dealing with parser frames?
4558          * The only extension currently using it is geoserver.
4559          *
4560          * @since 1.10
4561          * @todo better document or deprecate this
4562          *
4563          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4564          * @param $callback Mixed: the callback function (and object) to use for the tag
4565          * @return The old value of the mTagHooks array associated with the hook
4566          */
4567         function setTransparentTagHook( $tag, $callback ) {
4568                 $tag = strtolower( $tag );
4569                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4570                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4571                 $this->mTransparentTagHooks[$tag] = $callback;
4572
4573                 return $oldVal;
4574         }
4575
4576         /**
4577          * Remove all tag hooks
4578          */
4579         function clearTagHooks() {
4580                 $this->mTagHooks = array();
4581                 $this->mStripList = $this->mDefaultStripList;
4582         }
4583
4584         /**
4585          * Create a function, e.g. {{sum:1|2|3}}
4586          * The callback function should have the form:
4587          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4588          *
4589          * Or with SFH_OBJECT_ARGS:
4590          *    function myParserFunction( $parser, $frame, $args ) { ... }
4591          *
4592          * The callback may either return the text result of the function, or an array with the text
4593          * in element 0, and a number of flags in the other elements. The names of the flags are
4594          * specified in the keys. Valid flags are:
4595          *   found                     The text returned is valid, stop processing the template. This
4596          *                             is on by default.
4597          *   nowiki                    Wiki markup in the return value should be escaped
4598          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4599          *
4600          * @param $id String: The magic word ID
4601          * @param $callback Mixed: the callback function (and object) to use
4602          * @param $flags Integer: a combination of the following flags:
4603          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4604          *
4605          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4606          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4607          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4608          *     the arguments, and to control the way they are expanded.
4609          *
4610          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4611          *     arguments, for instance:
4612          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4613          *
4614          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4615          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4616          *     working if/when this is changed.
4617          *
4618          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4619          *     expansion.
4620          *
4621          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4622          *     about the methods available in PPFrame and PPNode.
4623          *
4624          * @return The old callback function for this name, if any
4625          */
4626         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4627                 global $wgContLang;
4628
4629                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4630                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4631
4632                 # Add to function cache
4633                 $mw = MagicWord::get( $id );
4634                 if ( !$mw )
4635                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4636
4637                 $synonyms = $mw->getSynonyms();
4638                 $sensitive = intval( $mw->isCaseSensitive() );
4639
4640                 foreach ( $synonyms as $syn ) {
4641                         # Case
4642                         if ( !$sensitive ) {
4643                                 $syn = $wgContLang->lc( $syn );
4644                         }
4645                         # Add leading hash
4646                         if ( !( $flags & SFH_NO_HASH ) ) {
4647                                 $syn = '#' . $syn;
4648                         }
4649                         # Remove trailing colon
4650                         if ( substr( $syn, -1, 1 ) === ':' ) {
4651                                 $syn = substr( $syn, 0, -1 );
4652                         }
4653                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4654                 }
4655                 return $oldVal;
4656         }
4657
4658         /**
4659          * Get all registered function hook identifiers
4660          *
4661          * @return Array
4662          */
4663         function getFunctionHooks() {
4664                 return array_keys( $this->mFunctionHooks );
4665         }
4666
4667         /**
4668          * Create a tag function, e.g. <test>some stuff</test>.
4669          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4670          * Unlike parser functions, their content is not preprocessed.
4671          */
4672         function setFunctionTagHook( $tag, $callback, $flags ) {
4673                 $tag = strtolower( $tag );
4674                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4675                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4676                         $this->mFunctionTagHooks[$tag] : null;
4677                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4678
4679                 if ( !in_array( $tag, $this->mStripList ) ) {
4680                         $this->mStripList[] = $tag;
4681                 }
4682
4683                 return $old;
4684         }
4685
4686         /**
4687          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4688          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4689          * Placeholders created in Skin::makeLinkObj()
4690          *
4691          * @param $text string
4692          * @param $options int
4693          *
4694          * @return array of link CSS classes, indexed by PDBK.
4695          */
4696         function replaceLinkHolders( &$text, $options = 0 ) {
4697                 return $this->mLinkHolders->replace( $text );
4698         }
4699
4700         /**
4701          * Replace <!--LINK--> link placeholders with plain text of links
4702          * (not HTML-formatted).
4703          *
4704          * @param $text String
4705          * @return String
4706          */
4707         function replaceLinkHoldersText( $text ) {
4708                 return $this->mLinkHolders->replaceText( $text );
4709         }
4710
4711         /**
4712          * Renders an image gallery from a text with one line per image.
4713          * text labels may be given by using |-style alternative text. E.g.
4714          *   Image:one.jpg|The number "1"
4715          *   Image:tree.jpg|A tree
4716          * given as text will return the HTML of a gallery with two images,
4717          * labeled 'The number "1"' and
4718          * 'A tree'.
4719          *
4720          * @param string $text
4721          * @param array $params
4722          * @return string HTML
4723          */
4724         function renderImageGallery( $text, $params ) {
4725                 $ig = new ImageGallery();
4726                 $ig->setContextTitle( $this->mTitle );
4727                 $ig->setShowBytes( false );
4728                 $ig->setShowFilename( false );
4729                 $ig->setParser( $this );
4730                 $ig->setHideBadImages();
4731                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4732
4733                 if ( isset( $params['showfilename'] ) ) {
4734                         $ig->setShowFilename( true );
4735                 } else {
4736                         $ig->setShowFilename( false );
4737                 }
4738                 if ( isset( $params['caption'] ) ) {
4739                         $caption = $params['caption'];
4740                         $caption = htmlspecialchars( $caption );
4741                         $caption = $this->replaceInternalLinks( $caption );
4742                         $ig->setCaptionHtml( $caption );
4743                 }
4744                 if ( isset( $params['perrow'] ) ) {
4745                         $ig->setPerRow( $params['perrow'] );
4746                 }
4747                 if ( isset( $params['widths'] ) ) {
4748                         $ig->setWidths( $params['widths'] );
4749                 }
4750                 if ( isset( $params['heights'] ) ) {
4751                         $ig->setHeights( $params['heights'] );
4752                 }
4753
4754                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4755
4756                 $lines = StringUtils::explode( "\n", $text );
4757                 foreach ( $lines as $line ) {
4758                         # match lines like these:
4759                         # Image:someimage.jpg|This is some image
4760                         $matches = array();
4761                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4762                         # Skip empty lines
4763                         if ( count( $matches ) == 0 ) {
4764                                 continue;
4765                         }
4766
4767                         if ( strpos( $matches[0], '%' ) !== false ) {
4768                                 $matches[1] = rawurldecode( $matches[1] );
4769                         }
4770                         $title = Title::newFromText( $matches[1], NS_FILE );
4771                         if ( is_null( $title ) ) {
4772                                 # Bogus title. Ignore these so we don't bomb out later.
4773                                 continue;
4774                         }
4775
4776                         $label = '';
4777                         $alt = '';
4778                         if ( isset( $matches[3] ) ) {
4779                                 // look for an |alt= definition while trying not to break existing
4780                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4781                                 // is defined for images in galleries
4782
4783                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4784                                 $altmatches = StringUtils::explode('|', $matches[3]);
4785                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4786
4787                                 foreach ( $altmatches as $altmatch ) {
4788                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4789                                         if ( $match ) {
4790                                                 $alt = $this->stripAltText( $match, false );
4791                                         }
4792                                         else {
4793                                                 // concatenate all other pipes
4794                                                 $label .= '|' . $altmatch;
4795                                         }
4796                                 }
4797                                 // remove the first pipe
4798                                 $label = substr( $label, 1 );
4799                         }
4800
4801                         $ig->add( $title, $label, $alt );
4802                 }
4803                 return $ig->toHTML();
4804         }
4805
4806         /**
4807          * @param $handler
4808          * @return array
4809          */
4810         function getImageParams( $handler ) {
4811                 if ( $handler ) {
4812                         $handlerClass = get_class( $handler );
4813                 } else {
4814                         $handlerClass = '';
4815                 }
4816                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4817                         # Initialise static lists
4818                         static $internalParamNames = array(
4819                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4820                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4821                                         'bottom', 'text-bottom' ),
4822                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4823                                         'upright', 'border', 'link', 'alt' ),
4824                         );
4825                         static $internalParamMap;
4826                         if ( !$internalParamMap ) {
4827                                 $internalParamMap = array();
4828                                 foreach ( $internalParamNames as $type => $names ) {
4829                                         foreach ( $names as $name ) {
4830                                                 $magicName = str_replace( '-', '_', "img_$name" );
4831                                                 $internalParamMap[$magicName] = array( $type, $name );
4832                                         }
4833                                 }
4834                         }
4835
4836                         # Add handler params
4837                         $paramMap = $internalParamMap;
4838                         if ( $handler ) {
4839                                 $handlerParamMap = $handler->getParamMap();
4840                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4841                                         $paramMap[$magic] = array( 'handler', $paramName );
4842                                 }
4843                         }
4844                         $this->mImageParams[$handlerClass] = $paramMap;
4845                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4846                 }
4847                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4848         }
4849
4850         /**
4851          * Parse image options text and use it to make an image
4852          *
4853          * @param $title Title
4854          * @param $options String
4855          * @param $holders LinkHolderArray|false
4856          * @return string HTML
4857          */
4858         function makeImage( $title, $options, $holders = false ) {
4859                 # Check if the options text is of the form "options|alt text"
4860                 # Options are:
4861                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4862                 #  * left       no resizing, just left align. label is used for alt= only
4863                 #  * right      same, but right aligned
4864                 #  * none       same, but not aligned
4865                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4866                 #  * center     center the image
4867                 #  * frame      Keep original image size, no magnify-button.
4868                 #  * framed     Same as "frame"
4869                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4870                 #  * upright    reduce width for upright images, rounded to full __0 px
4871                 #  * border     draw a 1px border around the image
4872                 #  * alt        Text for HTML alt attribute (defaults to empty)
4873                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4874                 # vertical-align values (no % or length right now):
4875                 #  * baseline
4876                 #  * sub
4877                 #  * super
4878                 #  * top
4879                 #  * text-top
4880                 #  * middle
4881                 #  * bottom
4882                 #  * text-bottom
4883
4884                 $parts = StringUtils::explode( "|", $options );
4885
4886                 # Give extensions a chance to select the file revision for us
4887                 $options = array();
4888                 $descQuery = false;
4889                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4890                         array( $this, $title, &$options, &$descQuery ) );
4891                 # Fetch and register the file (file title may be different via hooks)
4892                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
4893
4894                 # Get parameter map
4895                 $handler = $file ? $file->getHandler() : false;
4896
4897                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4898
4899                 if ( !$file ) {
4900                         $this->addTrackingCategory( 'broken-file-category' );
4901                 }
4902
4903                 # Process the input parameters
4904                 $caption = '';
4905                 $params = array( 'frame' => array(), 'handler' => array(),
4906                         'horizAlign' => array(), 'vertAlign' => array() );
4907                 foreach ( $parts as $part ) {
4908                         $part = trim( $part );
4909                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4910                         $validated = false;
4911                         if ( isset( $paramMap[$magicName] ) ) {
4912                                 list( $type, $paramName ) = $paramMap[$magicName];
4913
4914                                 # Special case; width and height come in one variable together
4915                                 if ( $type === 'handler' && $paramName === 'width' ) {
4916                                         $m = array();
4917                                         # (bug 13500) In both cases (width/height and width only),
4918                                         # permit trailing "px" for backward compatibility.
4919                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4920                                                 $width = intval( $m[1] );
4921                                                 $height = intval( $m[2] );
4922                                                 if ( $handler->validateParam( 'width', $width ) ) {
4923                                                         $params[$type]['width'] = $width;
4924                                                         $validated = true;
4925                                                 }
4926                                                 if ( $handler->validateParam( 'height', $height ) ) {
4927                                                         $params[$type]['height'] = $height;
4928                                                         $validated = true;
4929                                                 }
4930                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4931                                                 $width = intval( $value );
4932                                                 if ( $handler->validateParam( 'width', $width ) ) {
4933                                                         $params[$type]['width'] = $width;
4934                                                         $validated = true;
4935                                                 }
4936                                         } # else no validation -- bug 13436
4937                                 } else {
4938                                         if ( $type === 'handler' ) {
4939                                                 # Validate handler parameter
4940                                                 $validated = $handler->validateParam( $paramName, $value );
4941                                         } else {
4942                                                 # Validate internal parameters
4943                                                 switch( $paramName ) {
4944                                                 case 'manualthumb':
4945                                                 case 'alt':
4946                                                         # @todo FIXME: Possibly check validity here for
4947                                                         # manualthumb? downstream behavior seems odd with
4948                                                         # missing manual thumbs.
4949                                                         $validated = true;
4950                                                         $value = $this->stripAltText( $value, $holders );
4951                                                         break;
4952                                                 case 'link':
4953                                                         $chars = self::EXT_LINK_URL_CLASS;
4954                                                         $prots = $this->mUrlProtocols;
4955                                                         if ( $value === '' ) {
4956                                                                 $paramName = 'no-link';
4957                                                                 $value = true;
4958                                                                 $validated = true;
4959                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4960                                                                 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
4961                                                                         $paramName = 'link-url';
4962                                                                         $this->mOutput->addExternalLink( $value );
4963                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4964                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4965                                                                         }
4966                                                                         $validated = true;
4967                                                                 }
4968                                                         } else {
4969                                                                 $linkTitle = Title::newFromText( $value );
4970                                                                 if ( $linkTitle ) {
4971                                                                         $paramName = 'link-title';
4972                                                                         $value = $linkTitle;
4973                                                                         $this->mOutput->addLink( $linkTitle );
4974                                                                         $validated = true;
4975                                                                 }
4976                                                         }
4977                                                         break;
4978                                                 default:
4979                                                         # Most other things appear to be empty or numeric...
4980                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4981                                                 }
4982                                         }
4983
4984                                         if ( $validated ) {
4985                                                 $params[$type][$paramName] = $value;
4986                                         }
4987                                 }
4988                         }
4989                         if ( !$validated ) {
4990                                 $caption = $part;
4991                         }
4992                 }
4993
4994                 # Process alignment parameters
4995                 if ( $params['horizAlign'] ) {
4996                         $params['frame']['align'] = key( $params['horizAlign'] );
4997                 }
4998                 if ( $params['vertAlign'] ) {
4999                         $params['frame']['valign'] = key( $params['vertAlign'] );
5000                 }
5001
5002                 $params['frame']['caption'] = $caption;
5003
5004                 # Will the image be presented in a frame, with the caption below?
5005                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5006                                                  isset( $params['frame']['framed'] ) ||
5007                                                  isset( $params['frame']['thumbnail'] ) ||
5008                                                  isset( $params['frame']['manualthumb'] );
5009
5010                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5011                 # came to also set the caption, ordinary text after the image -- which
5012                 # makes no sense, because that just repeats the text multiple times in
5013                 # screen readers.  It *also* came to set the title attribute.
5014                 #
5015                 # Now that we have an alt attribute, we should not set the alt text to
5016                 # equal the caption: that's worse than useless, it just repeats the
5017                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5018                 # use the unnamed parameter for alt text as well, just for the time be-
5019                 # ing, if the unnamed param is set and the alt param is not.
5020                 #
5021                 # For the future, we need to figure out if we want to tweak this more,
5022                 # e.g., introducing a title= parameter for the title; ignoring the un-
5023                 # named parameter entirely for images without a caption; adding an ex-
5024                 # plicit caption= parameter and preserving the old magic unnamed para-
5025                 # meter for BC; ...
5026                 if ( $imageIsFramed ) { # Framed image
5027                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5028                                 # No caption or alt text, add the filename as the alt text so
5029                                 # that screen readers at least get some description of the image
5030                                 $params['frame']['alt'] = $title->getText();
5031                         }
5032                         # Do not set $params['frame']['title'] because tooltips don't make sense
5033                         # for framed images
5034                 } else { # Inline image
5035                         if ( !isset( $params['frame']['alt'] ) ) {
5036                                 # No alt text, use the "caption" for the alt text
5037                                 if ( $caption !== '') {
5038                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5039                                 } else {
5040                                         # No caption, fall back to using the filename for the
5041                                         # alt text
5042                                         $params['frame']['alt'] = $title->getText();
5043                                 }
5044                         }
5045                         # Use the "caption" for the tooltip text
5046                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5047                 }
5048
5049                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5050
5051                 # Linker does the rest
5052                 $time = isset( $options['time'] ) ? $options['time'] : false;
5053                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5054                         $time, $descQuery, $this->mOptions->getThumbSize() );
5055
5056                 # Give the handler a chance to modify the parser object
5057                 if ( $handler ) {
5058                         $handler->parserTransformHook( $this, $file );
5059                 }
5060
5061                 return $ret;
5062         }
5063
5064         /**
5065          * @param $caption
5066          * @param $holders LinkHolderArray
5067          * @return mixed|String
5068          */
5069         protected function stripAltText( $caption, $holders ) {
5070                 # Strip bad stuff out of the title (tooltip).  We can't just use
5071                 # replaceLinkHoldersText() here, because if this function is called
5072                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5073                 if ( $holders ) {
5074                         $tooltip = $holders->replaceText( $caption );
5075                 } else {
5076                         $tooltip = $this->replaceLinkHoldersText( $caption );
5077                 }
5078
5079                 # make sure there are no placeholders in thumbnail attributes
5080                 # that are later expanded to html- so expand them now and
5081                 # remove the tags
5082                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5083                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5084
5085                 return $tooltip;
5086         }
5087
5088         /**
5089          * Set a flag in the output object indicating that the content is dynamic and
5090          * shouldn't be cached.
5091          */
5092         function disableCache() {
5093                 wfDebug( "Parser output marked as uncacheable.\n" );
5094                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5095                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5096         }
5097
5098         /**
5099          * Callback from the Sanitizer for expanding items found in HTML attribute
5100          * values, so they can be safely tested and escaped.
5101          *
5102          * @param $text String
5103          * @param $frame PPFrame
5104          * @return String
5105          */
5106         function attributeStripCallback( &$text, $frame = false ) {
5107                 $text = $this->replaceVariables( $text, $frame );
5108                 $text = $this->mStripState->unstripBoth( $text );
5109                 return $text;
5110         }
5111
5112         /**
5113          * Accessor
5114          *
5115          * @return array
5116          */
5117         function getTags() {
5118                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5119         }
5120
5121         /**
5122          * Replace transparent tags in $text with the values given by the callbacks.
5123          *
5124          * Transparent tag hooks are like regular XML-style tag hooks, except they
5125          * operate late in the transformation sequence, on HTML instead of wikitext.
5126          *
5127          * @param $text string
5128          *
5129          * @return string
5130          */
5131         function replaceTransparentTags( $text ) {
5132                 $matches = array();
5133                 $elements = array_keys( $this->mTransparentTagHooks );
5134                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5135                 $replacements = array();
5136
5137                 foreach ( $matches as $marker => $data ) {
5138                         list( $element, $content, $params, $tag ) = $data;
5139                         $tagName = strtolower( $element );
5140                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5141                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5142                         } else {
5143                                 $output = $tag;
5144                         }
5145                         $replacements[$marker] = $output;
5146                 }
5147                 return strtr( $text, $replacements );
5148         }
5149
5150         /**
5151          * Break wikitext input into sections, and either pull or replace
5152          * some particular section's text.
5153          *
5154          * External callers should use the getSection and replaceSection methods.
5155          *
5156          * @param $text String: Page wikitext
5157          * @param $section String: a section identifier string of the form:
5158          *   <flag1> - <flag2> - ... - <section number>
5159          *
5160          * Currently the only recognised flag is "T", which means the target section number
5161          * was derived during a template inclusion parse, in other words this is a template
5162          * section edit link. If no flags are given, it was an ordinary section edit link.
5163          * This flag is required to avoid a section numbering mismatch when a section is
5164          * enclosed by <includeonly> (bug 6563).
5165          *
5166          * The section number 0 pulls the text before the first heading; other numbers will
5167          * pull the given section along with its lower-level subsections. If the section is
5168          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5169          *
5170          * Section 0 is always considered to exist, even if it only contains the empty
5171          * string. If $text is the empty string and section 0 is replaced, $newText is
5172          * returned.
5173          *
5174          * @param $mode String: one of "get" or "replace"
5175          * @param $newText String: replacement text for section data.
5176          * @return String: for "get", the extracted section text.
5177          *                 for "replace", the whole page with the section replaced.
5178          */
5179         private function extractSections( $text, $section, $mode, $newText='' ) {
5180                 global $wgTitle; # not generally used but removes an ugly failure mode
5181                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5182                 $outText = '';
5183                 $frame = $this->getPreprocessor()->newFrame();
5184
5185                 # Process section extraction flags
5186                 $flags = 0;
5187                 $sectionParts = explode( '-', $section );
5188                 $sectionIndex = array_pop( $sectionParts );
5189                 foreach ( $sectionParts as $part ) {
5190                         if ( $part === 'T' ) {
5191                                 $flags |= self::PTD_FOR_INCLUSION;
5192                         }
5193                 }
5194
5195                 # Check for empty input
5196                 if ( strval( $text ) === '' ) {
5197                         # Only sections 0 and T-0 exist in an empty document
5198                         if ( $sectionIndex == 0 ) {
5199                                 if ( $mode === 'get' ) {
5200                                         return '';
5201                                 } else {
5202                                         return $newText;
5203                                 }
5204                         } else {
5205                                 if ( $mode === 'get' ) {
5206                                         return $newText;
5207                                 } else {
5208                                         return $text;
5209                                 }
5210                         }
5211                 }
5212
5213                 # Preprocess the text
5214                 $root = $this->preprocessToDom( $text, $flags );
5215
5216                 # <h> nodes indicate section breaks
5217                 # They can only occur at the top level, so we can find them by iterating the root's children
5218                 $node = $root->getFirstChild();
5219
5220                 # Find the target section
5221                 if ( $sectionIndex == 0 ) {
5222                         # Section zero doesn't nest, level=big
5223                         $targetLevel = 1000;
5224                 } else {
5225                         while ( $node ) {
5226                                 if ( $node->getName() === 'h' ) {
5227                                         $bits = $node->splitHeading();
5228                                         if ( $bits['i'] == $sectionIndex ) {
5229                                                 $targetLevel = $bits['level'];
5230                                                 break;
5231                                         }
5232                                 }
5233                                 if ( $mode === 'replace' ) {
5234                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5235                                 }
5236                                 $node = $node->getNextSibling();
5237                         }
5238                 }
5239
5240                 if ( !$node ) {
5241                         # Not found
5242                         if ( $mode === 'get' ) {
5243                                 return $newText;
5244                         } else {
5245                                 return $text;
5246                         }
5247                 }
5248
5249                 # Find the end of the section, including nested sections
5250                 do {
5251                         if ( $node->getName() === 'h' ) {
5252                                 $bits = $node->splitHeading();
5253                                 $curLevel = $bits['level'];
5254                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5255                                         break;
5256                                 }
5257                         }
5258                         if ( $mode === 'get' ) {
5259                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5260                         }
5261                         $node = $node->getNextSibling();
5262                 } while ( $node );
5263
5264                 # Write out the remainder (in replace mode only)
5265                 if ( $mode === 'replace' ) {
5266                         # Output the replacement text
5267                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5268                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5269                         # Only add trailing whitespace if there is newText
5270                         if ( $newText != "" ) {
5271                                 $outText .= $newText . "\n\n";
5272                         }
5273
5274                         while ( $node ) {
5275                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5276                                 $node = $node->getNextSibling();
5277                         }
5278                 }
5279
5280                 if ( is_string( $outText ) ) {
5281                         # Re-insert stripped tags
5282                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5283                 }
5284
5285                 return $outText;
5286         }
5287
5288         /**
5289          * This function returns the text of a section, specified by a number ($section).
5290          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5291          * the first section before any such heading (section 0).
5292          *
5293          * If a section contains subsections, these are also returned.
5294          *
5295          * @param $text String: text to look in
5296          * @param $section String: section identifier
5297          * @param $deftext String: default to return if section is not found
5298          * @return string text of the requested section
5299          */
5300         public function getSection( $text, $section, $deftext='' ) {
5301                 return $this->extractSections( $text, $section, "get", $deftext );
5302         }
5303
5304         /**
5305          * This function returns $oldtext after the content of the section
5306          * specified by $section has been replaced with $text. If the target
5307          * section does not exist, $oldtext is returned unchanged.
5308          *
5309          * @param $oldtext String: former text of the article
5310          * @param $section Numeric: section identifier
5311          * @param $text String: replacing text
5312          * @return String: modified text
5313          */
5314         public function replaceSection( $oldtext, $section, $text ) {
5315                 return $this->extractSections( $oldtext, $section, "replace", $text );
5316         }
5317
5318         /**
5319          * Get the ID of the revision we are parsing
5320          *
5321          * @return Mixed: integer or null
5322          */
5323         function getRevisionId() {
5324                 return $this->mRevisionId;
5325         }
5326
5327         /**
5328          * Get the revision object for $this->mRevisionId
5329          *
5330          * @return Revision|null either a Revision object or null
5331          */
5332         protected function getRevisionObject() {
5333                 if ( !is_null( $this->mRevisionObject ) ) {
5334                         return $this->mRevisionObject;
5335                 }
5336                 if ( is_null( $this->mRevisionId ) ) {
5337                         return null;
5338                 }
5339
5340                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5341                 return $this->mRevisionObject;
5342         }
5343
5344         /**
5345          * Get the timestamp associated with the current revision, adjusted for
5346          * the default server-local timestamp
5347          */
5348         function getRevisionTimestamp() {
5349                 if ( is_null( $this->mRevisionTimestamp ) ) {
5350                         wfProfileIn( __METHOD__ );
5351
5352                         $revObject = $this->getRevisionObject();
5353                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5354
5355                         if( $timestamp !== false ) {
5356                                 global $wgContLang;
5357
5358                                 # The cryptic '' timezone parameter tells to use the site-default
5359                                 # timezone offset instead of the user settings.
5360                                 #
5361                                 # Since this value will be saved into the parser cache, served
5362                                 # to other users, and potentially even used inside links and such,
5363                                 # it needs to be consistent for all visitors.
5364                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5365                         }
5366
5367                         wfProfileOut( __METHOD__ );
5368                 }
5369                 return $this->mRevisionTimestamp;
5370         }
5371
5372         /**
5373          * Get the name of the user that edited the last revision
5374          *
5375          * @return String: user name
5376          */
5377         function getRevisionUser() {
5378                 if( is_null( $this->mRevisionUser ) ) {
5379                         $revObject = $this->getRevisionObject();
5380
5381                         # if this template is subst: the revision id will be blank,
5382                         # so just use the current user's name
5383                         if( $revObject ) {
5384                                 $this->mRevisionUser = $revObject->getUserText();
5385                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5386                                 $this->mRevisionUser = $this->getUser()->getName();
5387                         }
5388                 }
5389                 return $this->mRevisionUser;
5390         }
5391
5392         /**
5393          * Mutator for $mDefaultSort
5394          *
5395          * @param $sort New value
5396          */
5397         public function setDefaultSort( $sort ) {
5398                 $this->mDefaultSort = $sort;
5399                 $this->mOutput->setProperty( 'defaultsort', $sort );
5400         }
5401
5402         /**
5403          * Accessor for $mDefaultSort
5404          * Will use the empty string if none is set.
5405          *
5406          * This value is treated as a prefix, so the
5407          * empty string is equivalent to sorting by
5408          * page name.
5409          *
5410          * @return string
5411          */
5412         public function getDefaultSort() {
5413                 if ( $this->mDefaultSort !== false ) {
5414                         return $this->mDefaultSort;
5415                 } else {
5416                         return '';
5417                 }
5418         }
5419
5420         /**
5421          * Accessor for $mDefaultSort
5422          * Unlike getDefaultSort(), will return false if none is set
5423          *
5424          * @return string or false
5425          */
5426         public function getCustomDefaultSort() {
5427                 return $this->mDefaultSort;
5428         }
5429
5430         /**
5431          * Try to guess the section anchor name based on a wikitext fragment
5432          * presumably extracted from a heading, for example "Header" from
5433          * "== Header ==".
5434          *
5435          * @param $text string
5436          *
5437          * @return string
5438          */
5439         public function guessSectionNameFromWikiText( $text ) {
5440                 # Strip out wikitext links(they break the anchor)
5441                 $text = $this->stripSectionName( $text );
5442                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5443                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5444         }
5445
5446         /**
5447          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5448          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5449          * as something other than UTF-8 (apparently?), resulting in breakage.
5450          *
5451          * @param $text String: The section name
5452          * @return string An anchor
5453          */
5454         public function guessLegacySectionNameFromWikiText( $text ) {
5455                 # Strip out wikitext links(they break the anchor)
5456                 $text = $this->stripSectionName( $text );
5457                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5458                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5459         }
5460
5461         /**
5462          * Strips a text string of wikitext for use in a section anchor
5463          *
5464          * Accepts a text string and then removes all wikitext from the
5465          * string and leaves only the resultant text (i.e. the result of
5466          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5467          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5468          * to create valid section anchors by mimicing the output of the
5469          * parser when headings are parsed.
5470          *
5471          * @param $text String: text string to be stripped of wikitext
5472          * for use in a Section anchor
5473          * @return Filtered text string
5474          */
5475         public function stripSectionName( $text ) {
5476                 # Strip internal link markup
5477                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5478                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5479
5480                 # Strip external link markup
5481                 # @todo FIXME: Not tolerant to blank link text
5482                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5483                 # on how many empty links there are on the page - need to figure that out.
5484                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5485
5486                 # Parse wikitext quotes (italics & bold)
5487                 $text = $this->doQuotes( $text );
5488
5489                 # Strip HTML tags
5490                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5491                 return $text;
5492         }
5493
5494         /**
5495          * strip/replaceVariables/unstrip for preprocessor regression testing
5496          *
5497          * @param $text string
5498          * @param $title Title
5499          * @param $options ParserOptions
5500          * @param $outputType int
5501          *
5502          * @return string
5503          */
5504         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5505                 $this->startParse( $title, $options, $outputType, true );
5506
5507                 $text = $this->replaceVariables( $text );
5508                 $text = $this->mStripState->unstripBoth( $text );
5509                 $text = Sanitizer::removeHTMLtags( $text );
5510                 return $text;
5511         }
5512
5513         /**
5514          * @param $text string
5515          * @param $title Title
5516          * @param $options ParserOptions
5517          * @return string
5518          */
5519         function testPst( $text, Title $title, ParserOptions $options ) {
5520                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5521         }
5522
5523         /**
5524          * @param $text
5525          * @param $title Title
5526          * @param $options ParserOptions
5527          * @return string
5528          */
5529         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5530                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5531         }
5532
5533         /**
5534          * Call a callback function on all regions of the given text that are not
5535          * inside strip markers, and replace those regions with the return value
5536          * of the callback. For example, with input:
5537          *
5538          *  aaa<MARKER>bbb
5539          *
5540          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5541          * two strings will be replaced with the value returned by the callback in
5542          * each case.
5543          *
5544          * @param $s string
5545          * @param $callback
5546          *
5547          * @return string
5548          */
5549         function markerSkipCallback( $s, $callback ) {
5550                 $i = 0;
5551                 $out = '';
5552                 while ( $i < strlen( $s ) ) {
5553                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5554                         if ( $markerStart === false ) {
5555                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5556                                 break;
5557                         } else {
5558                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5559                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5560                                 if ( $markerEnd === false ) {
5561                                         $out .= substr( $s, $markerStart );
5562                                         break;
5563                                 } else {
5564                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5565                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5566                                         $i = $markerEnd;
5567                                 }
5568                         }
5569                 }
5570                 return $out;
5571         }
5572
5573         /**
5574          * Save the parser state required to convert the given half-parsed text to
5575          * HTML. "Half-parsed" in this context means the output of
5576          * recursiveTagParse() or internalParse(). This output has strip markers
5577          * from replaceVariables (extensionSubstitution() etc.), and link
5578          * placeholders from replaceLinkHolders().
5579          *
5580          * Returns an array which can be serialized and stored persistently. This
5581          * array can later be loaded into another parser instance with
5582          * unserializeHalfParsedText(). The text can then be safely incorporated into
5583          * the return value of a parser hook.
5584          *
5585          * @param $text string
5586          *
5587          * @return array
5588          */
5589         function serializeHalfParsedText( $text ) {
5590                 wfProfileIn( __METHOD__ );
5591                 $data = array(
5592                         'text' => $text,
5593                         'version' => self::HALF_PARSED_VERSION,
5594                         'stripState' => $this->mStripState->getSubState( $text ),
5595                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5596                 );
5597                 wfProfileOut( __METHOD__ );
5598                 return $data;
5599         }
5600
5601         /**
5602          * Load the parser state given in the $data array, which is assumed to
5603          * have been generated by serializeHalfParsedText(). The text contents is
5604          * extracted from the array, and its markers are transformed into markers
5605          * appropriate for the current Parser instance. This transformed text is
5606          * returned, and can be safely included in the return value of a parser
5607          * hook.
5608          *
5609          * If the $data array has been stored persistently, the caller should first
5610          * check whether it is still valid, by calling isValidHalfParsedText().
5611          *
5612          * @param $data Serialized data
5613          * @return String
5614          */
5615         function unserializeHalfParsedText( $data ) {
5616                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5617                         throw new MWException( __METHOD__.': invalid version' );
5618                 }
5619
5620                 # First, extract the strip state.
5621                 $texts = array( $data['text'] );
5622                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5623
5624                 # Now renumber links
5625                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5626
5627                 # Should be good to go.
5628                 return $texts[0];
5629         }
5630
5631         /**
5632          * Returns true if the given array, presumed to be generated by
5633          * serializeHalfParsedText(), is compatible with the current version of the
5634          * parser.
5635          *
5636          * @param $data Array
5637          *
5638          * @return bool
5639          */
5640         function isValidHalfParsedText( $data ) {
5641                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5642         }
5643 }