includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *     produces HTML output
  21  * preSaveTransform().
  22  *     produces altered wiki markup.
  23  * preprocess()
  24  *     removes HTML comments and expands templates
  25  * cleanSig() / cleanSigInSig()
  26  *     Cleans a signature before saving it to preferences
  27  * getSection()
  28  *     Return the content of a section from an article for section editing
  29  * replaceSection()
  30  *     Replaces a section by number inside an article
  31  * getPreloadText()
  32  *     Removes <noinclude> sections, and <includeonly> tags.
  33  *
  34  * Globals used:
  35  *    objects:   $wgLang, $wgContLang
  36  *
  37  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.4';
  57
  58         # Flags for Parser::setFunctionHook
  59         # Also available as global constants from Defines.php
  60         const SFH_NO_HASH = 1;
  61         const SFH_OBJECT_ARGS = 2;
  62
  63         # Constants needed for external link processing
  64         # Everything except bracket, space, or control characters
  65         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  66         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  67                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  68
  69         # State constants for the definition list colon extraction
  70         const COLON_STATE_TEXT = 0;
  71         const COLON_STATE_TAG = 1;
  72         const COLON_STATE_TAGSTART = 2;
  73         const COLON_STATE_CLOSETAG = 3;
  74         const COLON_STATE_TAGSLASH = 4;
  75         const COLON_STATE_COMMENT = 5;
  76         const COLON_STATE_COMMENTDASH = 6;
  77         const COLON_STATE_COMMENTDASHDASH = 7;
  78
  79         # Flags for preprocessToDom
  80         const PTD_FOR_INCLUSION = 1;
  81
  82         # Allowed values for $this->mOutputType
  83         # Parameter to startExternalParse().
  84         const OT_HTML = 1; # like parse()
  85         const OT_WIKI = 2; # like preSaveTransform()
  86         const OT_PREPROCESS = 3; # like preprocess()
  87         const OT_MSG = 3;
  88         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  89
  90         # Marker Suffix needs to be accessible staticly.
  91         const MARKER_SUFFIX = "-QINU\x7f";
  92
  93         # Persistent:
  94         var $mTagHooks = array();
  95         var $mTransparentTagHooks = array();
  96         var $mFunctionHooks = array();
  97         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
  98         var $mFunctionTagHooks = array();
  99         var $mStripList  = array();
 100         var $mDefaultStripList  = array();
 101         var $mVarCache = array();
 102         var $mImageParams = array();
 103         var $mImageParamsMagicArray = array();
 104         var $mMarkerIndex = 0;
 105         var $mFirstCall = true;
 106         var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
 107         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 108
 109
 110         # Cleared with clearState():
 111         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 112         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 113         var $mLinkHolders, $mLinkID;
 114         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 115         var $mTplExpandCache; # empty-frame expansion cache
 116         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 117         var $mExpensiveFunctionCount; # number of expensive parser function calls
 118         var $mUser; # User object; only used when doing pre-save transform
 119
 120         # Temporary
 121         # These are variables reset at least once per parse regardless of $clearState
 122         var $mOptions;      # ParserOptions object
 123         var $mTitle;        # Title context, used for self-link rendering and similar things
 124         var $mOutputType;   # Output type, one of the OT_xxx constants
 125         var $ot;            # Shortcut alias, see setOutputType()
 126         var $mRevisionObject; # The revision object of the specified revision ID
 127         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 128         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 129         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 130         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 131
 132         /**
 133          * Constructor
 134          *
 135          * @public
 136          */
 137         function __construct( $conf = array() ) {
 138                 $this->mConf = $conf;
 139                 $this->mUrlProtocols = wfUrlProtocols();
 140                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 141                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 142                 if ( isset( $conf['preprocessorClass'] ) ) {
 143                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 144                 } elseif ( extension_loaded( 'domxml' ) ) {
 145                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 146                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 147                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 148                 } elseif ( extension_loaded( 'dom' ) ) {
 149                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 150                 } else {
 151                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 152                 }
 153         }
 154
 155         /**
 156          * Reduce memory usage to reduce the impact of circular references
 157          */
 158         function __destruct() {
 159                 if ( isset( $this->mLinkHolders ) ) {
 160                         $this->mLinkHolders->__destruct();
 161                 }
 162                 foreach ( $this as $name => $value ) {
 163                         unset( $this->$name );
 164                 }
 165         }
 166
 167         /**
 168          * Do various kinds of initialisation on the first call of the parser
 169          */
 170         function firstCallInit() {
 171                 if ( !$this->mFirstCall ) {
 172                         return;
 173                 }
 174                 $this->mFirstCall = false;
 175
 176                 wfProfileIn( __METHOD__ );
 177
 178                 CoreParserFunctions::register( $this );
 179                 CoreTagHooks::register( $this );
 180                 $this->initialiseVariables();
 181
 182                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 183                 wfProfileOut( __METHOD__ );
 184         }
 185
 186         /**
 187          * Clear Parser state
 188          *
 189          * @private
 190          */
 191         function clearState() {
 192                 wfProfileIn( __METHOD__ );
 193                 if ( $this->mFirstCall ) {
 194                         $this->firstCallInit();
 195                 }
 196                 $this->mOutput = new ParserOutput;
 197                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 198                 $this->mAutonumber = 0;
 199                 $this->mLastSection = '';
 200                 $this->mDTopen = false;
 201                 $this->mIncludeCount = array();
 202                 $this->mStripState = new StripState;
 203                 $this->mArgStack = false;
 204                 $this->mInPre = false;
 205                 $this->mLinkHolders = new LinkHolderArray( $this );
 206                 $this->mLinkID = 0;
 207                 $this->mRevisionObject = $this->mRevisionTimestamp =
 208                         $this->mRevisionId = $this->mRevisionUser = null;
 209                 $this->mVarCache = array();
 210                 $this->mUser = null;
 211
 212                 /**
 213                  * Prefix for temporary replacement strings for the multipass parser.
 214                  * \x07 should never appear in input as it's disallowed in XML.
 215                  * Using it at the front also gives us a little extra robustness
 216                  * since it shouldn't match when butted up against identifier-like
 217                  * string constructs.
 218                  *
 219                  * Must not consist of all title characters, or else it will change
 220                  * the behaviour of <nowiki> in a link.
 221                  */
 222                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 223                 # Changed to \x7f to allow XML double-parsing -- TS
 224                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 225
 226
 227                 # Clear these on every parse, bug 4549
 228                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 229
 230                 $this->mShowToc = true;
 231                 $this->mForceTocPosition = false;
 232                 $this->mIncludeSizes = array(
 233                         'post-expand' => 0,
 234                         'arg' => 0,
 235                 );
 236                 $this->mPPNodeCount = 0;
 237                 $this->mDefaultSort = false;
 238                 $this->mHeadings = array();
 239                 $this->mDoubleUnderscores = array();
 240                 $this->mExpensiveFunctionCount = 0;
 241
 242                 # Fix cloning
 243                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 244                         $this->mPreprocessor = null;
 245                 }
 246
 247                 wfRunHooks( 'ParserClearState', array( &$this ) );
 248                 wfProfileOut( __METHOD__ );
 249         }
 250
 251         /**
 252          * Convert wikitext to HTML
 253          * Do not call this function recursively.
 254          *
 255          * @param $text String: text we want to parse
 256          * @param $title A title object
 257          * @param $options ParserOptions
 258          * @param $linestart boolean
 259          * @param $clearState boolean
 260          * @param $revid Int: number to pass in {{REVISIONID}}
 261          * @return ParserOutput a ParserOutput
 262          */
 263         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 264                 /**
 265                  * First pass--just handle <nowiki> sections, pass the rest off
 266                  * to internalParse() which does all the real work.
 267                  */
 268
 269                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 270                 $fname = __METHOD__.'-' . wfGetCaller();
 271                 wfProfileIn( __METHOD__ );
 272                 wfProfileIn( $fname );
 273
 274                 $this->startExternalParse( $title, $options, self::OT_HTML, $clearState );
 275
 276                 $oldRevisionId = $this->mRevisionId;
 277                 $oldRevisionObject = $this->mRevisionObject;
 278                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 279                 $oldRevisionUser = $this->mRevisionUser;
 280                 if ( $revid !== null ) {
 281                         $this->mRevisionId = $revid;
 282                         $this->mRevisionObject = null;
 283                         $this->mRevisionTimestamp = null;
 284                         $this->mRevisionUser = null;
 285                 }
 286
 287                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 288                 # No more strip!
 289                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 290                 $text = $this->internalParse( $text );
 291
 292                 $text = $this->mStripState->unstripGeneral( $text );
 293
 294                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 295                 $fixtags = array(
 296                         # french spaces, last one Guillemet-left
 297                         # only if there is something before the space
 298                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 299                         # french spaces, Guillemet-right
 300                         '/(\\302\\253) /' => '\\1&#160;',
 301                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 302                 );
 303                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 304
 305                 $text = $this->doBlockLevels( $text, $linestart );
 306
 307                 $this->replaceLinkHolders( $text );
 308
 309                 /**
 310                  * The page doesn't get language converted if
 311                  * a) It's disabled
 312                  * b) Content isn't converted
 313                  * c) It's a conversion table
 314                  */
 315                 if ( !( $wgDisableLangConversion
 316                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 317                                 || $this->mTitle->isConversionTable() ) ) {
 318
 319                         # The position of the convert() call should not be changed. it
 320                         # assumes that the links are all replaced and the only thing left
 321                         # is the <nowiki> mark.
 322
 323                         $text = $wgContLang->convert( $text );
 324                 }
 325
 326                 /**
 327                  * A converted title will be provided in the output object if title and
 328                  * content conversion are enabled, the article text does not contain
 329                  * a conversion-suppressing double-underscore tag, and no
 330                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 331                  * automatic link conversion.
 332                  */
 333                 if ( !( $wgDisableLangConversion
 334                                 || $wgDisableTitleConversion
 335                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 336                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 337                                 || $this->mOutput->getDisplayTitle() !== false ) )
 338                 {
 339                         $convruletitle = $wgContLang->getConvRuleTitle();
 340                         if ( $convruletitle ) {
 341                                 $this->mOutput->setTitleText( $convruletitle );
 342                         } else {
 343                                 $titleText = $wgContLang->convertTitle( $title );
 344                                 $this->mOutput->setTitleText( $titleText );
 345                         }
 346                 }
 347
 348                 $text = $this->mStripState->unstripNoWiki( $text );
 349
 350                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 351
 352 //!JF Move to its own function
 353
 354                 $uniq_prefix = $this->mUniqPrefix;
 355                 $matches = array();
 356                 $elements = array_keys( $this->mTransparentTagHooks );
 357                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 358
 359                 foreach ( $matches as $marker => $data ) {
 360                         list( $element, $content, $params, $tag ) = $data;
 361                         $tagName = strtolower( $element );
 362                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 363                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 364                         } else {
 365                                 $output = $tag;
 366                         }
 367                         $this->mStripState->general->setPair( $marker, $output );
 368                 }
 369                 $text = $this->mStripState->unstripGeneral( $text );
 370
 371                 $text = Sanitizer::normalizeCharReferences( $text );
 372
 373                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 374                         $text = MWTidy::tidy( $text );
 375                 } else {
 376                         # attempt to sanitize at least some nesting problems
 377                         # (bug #2702 and quite a few others)
 378                         $tidyregs = array(
 379                                 # ''Something [http://www.cool.com cool''] -->
 380                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 381                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 382                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 383                                 # fix up an anchor inside another anchor, only
 384                                 # at least for a single single nested link (bug 3695)
 385                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 386                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 387                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 388                                 # contains a div, so fix it up here; replace
 389                                 # div with escaped text
 390                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 391                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 392                                 # remove empty italic or bold tag pairs, some
 393                                 # introduced by rules above
 394                                 '/<([bi])><\/\\1>/' => '',
 395                         );
 396
 397                         $text = preg_replace(
 398                                 array_keys( $tidyregs ),
 399                                 array_values( $tidyregs ),
 400                                 $text );
 401                 }
 402                 global $wgExpensiveParserFunctionLimit;
 403                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 404                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 405                 }
 406
 407                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 408
 409                 # Information on include size limits, for the benefit of users who try to skirt them
 410                 if ( $this->mOptions->getEnableLimitReport() ) {
 411                         $max = $this->mOptions->getMaxIncludeSize();
 412                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 413                         $limitReport =
 414                                 "NewPP limit report\n" .
 415                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 416                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 417                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 418                                 $PFreport;
 419                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 420                         $text .= "\n<!-- \n$limitReport-->\n";
 421                 }
 422                 $this->mOutput->setText( $text );
 423
 424                 $this->mRevisionId = $oldRevisionId;
 425                 $this->mRevisionObject = $oldRevisionObject;
 426                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 427                 $this->mRevisionUser = $oldRevisionUser;
 428                 wfProfileOut( $fname );
 429                 wfProfileOut( __METHOD__ );
 430
 431                 return $this->mOutput;
 432         }
 433
 434         /**
 435          * Recursive parser entry point that can be called from an extension tag
 436          * hook.
 437          *
 438          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 439          *
 440          * @param $text String: text extension wants to have parsed
 441          * @param $frame PPFrame: The frame to use for expanding any template variables
 442          */
 443         function recursiveTagParse( $text, $frame=false ) {
 444                 wfProfileIn( __METHOD__ );
 445                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 446                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 447                 $text = $this->internalParse( $text, false, $frame );
 448                 wfProfileOut( __METHOD__ );
 449                 return $text;
 450         }
 451
 452         /**
 453          * Expand templates and variables in the text, producing valid, static wikitext.
 454          * Also removes comments.
 455          */
 456         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 457                 wfProfileIn( __METHOD__ );
 458                 $this->startExternalParse( $title, $options, self::OT_PREPROCESS, true );
 459                 if ( $revid !== null ) {
 460                         $this->mRevisionId = $revid;
 461                 }
 462                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 463                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 464                 $text = $this->replaceVariables( $text );
 465                 $text = $this->mStripState->unstripBoth( $text );
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Process the wikitext for the ?preload= feature. (bug 5210)
 472          *
 473          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 474          * comments, templates, arguments, tags hooks and parser functions are untouched.
 475          */
 476         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 477                 # Parser (re)initialisation
 478                 $this->startExternalParse( $title, $options, self::OT_PLAIN, true );
 479
 480                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 481                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 482                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 483         }
 484
 485         /**
 486          * Get a random string
 487          *
 488          * @private
 489          * @static
 490          */
 491         static private function getRandomString() {
 492                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 493         }
 494
 495         /**
 496          * Set the current user.
 497          * Should only be used when doing pre-save transform.
 498          *
 499          * @param $user Mixed: User object or null (to reset)
 500          */
 501         function setUser( $user ) {
 502                 $this->mUser = $user;
 503         }
 504
 505         /**
 506          * Accessor for mUniqPrefix.
 507          *
 508          * @return String
 509          */
 510         public function uniqPrefix() {
 511                 if ( !isset( $this->mUniqPrefix ) ) {
 512                         # @todo Fixme: this is probably *horribly wrong*
 513                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 514                         # if this is called for a parser cache hit, the parser may not
 515                         # have ever been initialized in the first place.
 516                         # Not really sure what the heck is supposed to be going on here.
 517                         return '';
 518                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 519                 }
 520                 return $this->mUniqPrefix;
 521         }
 522
 523         /**
 524          * Set the context title
 525          */
 526         function setTitle( $t ) {
 527                 if ( !$t || $t instanceof FakeTitle ) {
 528                         $t = Title::newFromText( 'NO TITLE' );
 529                 }
 530
 531                 if ( strval( $t->getFragment() ) !== '' ) {
 532                         # Strip the fragment to avoid various odd effects
 533                         $this->mTitle = clone $t;
 534                         $this->mTitle->setFragment( '' );
 535                 } else {
 536                         $this->mTitle = $t;
 537                 }
 538         }
 539
 540         /**
 541          * Accessor for the Title object
 542          *
 543          * @return Title object
 544          */
 545         function getTitle() {
 546                 return $this->mTitle;
 547         }
 548
 549         /**
 550          * Accessor/mutator for the Title object
 551          *
 552          * @param $x New Title object or null to just get the current one
 553          * @return Title object
 554          */
 555         function Title( $x = null ) {
 556                 return wfSetVar( $this->mTitle, $x );
 557         }
 558
 559         /**
 560          * Set the output type
 561          *
 562          * @param $ot Integer: new value
 563          */
 564         function setOutputType( $ot ) {
 565                 $this->mOutputType = $ot;
 566                 # Shortcut alias
 567                 $this->ot = array(
 568                         'html' => $ot == self::OT_HTML,
 569                         'wiki' => $ot == self::OT_WIKI,
 570                         'pre' => $ot == self::OT_PREPROCESS,
 571                         'plain' => $ot == self::OT_PLAIN,
 572                 );
 573         }
 574
 575         /**
 576          * Accessor/mutator for the output type
 577          *
 578          * @param $x New value or null to just get the current one
 579          * @return Integer
 580          */
 581         function OutputType( $x = null ) {
 582                 return wfSetVar( $this->mOutputType, $x );
 583         }
 584
 585         /**
 586          * Get the ParserOutput object
 587          *
 588          * @return ParserOutput object
 589          */
 590         function getOutput() {
 591                 return $this->mOutput;
 592         }
 593
 594         /**
 595          * Get the ParserOptions object
 596          *
 597          * @return ParserOptions object
 598          */
 599         function getOptions() {
 600                 return $this->mOptions;
 601         }
 602
 603         /**
 604          * Accessor/mutator for the ParserOptions object
 605          *
 606          * @param $x New value or null to just get the current one
 607          * @return Current ParserOptions object
 608          */
 609         function Options( $x = null ) {
 610                 return wfSetVar( $this->mOptions, $x );
 611         }
 612
 613         function nextLinkID() {
 614                 return $this->mLinkID++;
 615         }
 616
 617         function getFunctionLang() {
 618                 global $wgLang, $wgContLang;
 619
 620                 $target = $this->mOptions->getTargetLanguage();
 621                 if ( $target !== null ) {
 622                         return $target;
 623                 } else {
 624                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 625                 }
 626         }
 627
 628         /**
 629          * Get a User object either from $this->mUser, if set, or from the
 630          * ParserOptions object otherwise
 631          *
 632          * @return User object
 633          */
 634         function getUser() {
 635                 if ( !is_null( $this->mUser ) ) {
 636                         return $this->mUser;
 637                 }
 638                 return $this->mOptions->getUser();
 639         }
 640
 641         /**
 642          * Get a preprocessor object
 643          *
 644          * @return Preprocessor instance
 645          */
 646         function getPreprocessor() {
 647                 if ( !isset( $this->mPreprocessor ) ) {
 648                         $class = $this->mPreprocessorClass;
 649                         $this->mPreprocessor = new $class( $this );
 650                 }
 651                 return $this->mPreprocessor;
 652         }
 653
 654         /**
 655          * Replaces all occurrences of HTML-style comments and the given tags
 656          * in the text with a random marker and returns the next text. The output
 657          * parameter $matches will be an associative array filled with data in
 658          * the form:
 659          *   'UNIQ-xxxxx' => array(
 660          *     'element',
 661          *     'tag content',
 662          *     array( 'param' => 'x' ),
 663          *     '<element param="x">tag content</element>' ) )
 664          *
 665          * @param $elements list of element names. Comments are always extracted.
 666          * @param $text Source text string.
 667          * @param $matches Out parameter, Array: extracted tags
 668          * @param $uniq_prefix
 669          * @return String: stripped text
 670          *
 671          * @static
 672          */
 673         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 674                 static $n = 1;
 675                 $stripped = '';
 676                 $matches = array();
 677
 678                 $taglist = implode( '|', $elements );
 679                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 680
 681                 while ( $text != '' ) {
 682                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 683                         $stripped .= $p[0];
 684                         if ( count( $p ) < 5 ) {
 685                                 break;
 686                         }
 687                         if ( count( $p ) > 5 ) {
 688                                 # comment
 689                                 $element    = $p[4];
 690                                 $attributes = '';
 691                                 $close      = '';
 692                                 $inside     = $p[5];
 693                         } else {
 694                                 # tag
 695                                 $element    = $p[1];
 696                                 $attributes = $p[2];
 697                                 $close      = $p[3];
 698                                 $inside     = $p[4];
 699                         }
 700
 701                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 702                         $stripped .= $marker;
 703
 704                         if ( $close === '/>' ) {
 705                                 # Empty element tag, <tag />
 706                                 $content = null;
 707                                 $text = $inside;
 708                                 $tail = null;
 709                         } else {
 710                                 if ( $element === '!--' ) {
 711                                         $end = '/(-->)/';
 712                                 } else {
 713                                         $end = "/(<\\/$element\\s*>)/i";
 714                                 }
 715                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 716                                 $content = $q[0];
 717                                 if ( count( $q ) < 3 ) {
 718                                         # No end tag -- let it run out to the end of the text.
 719                                         $tail = '';
 720                                         $text = '';
 721                                 } else {
 722                                         $tail = $q[1];
 723                                         $text = $q[2];
 724                                 }
 725                         }
 726
 727                         $matches[$marker] = array( $element,
 728                                 $content,
 729                                 Sanitizer::decodeTagAttributes( $attributes ),
 730                                 "<$element$attributes$close$content$tail" );
 731                 }
 732                 return $stripped;
 733         }
 734
 735         /**
 736          * Get a list of strippable XML-like elements
 737          */
 738         function getStripList() {
 739                 return $this->mStripList;
 740         }
 741
 742         /**
 743          * @deprecated use replaceVariables
 744          */
 745         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 746                 return $text;
 747         }
 748
 749         /**
 750          * Restores pre, math, and other extensions removed by strip()
 751          *
 752          * always call unstripNoWiki() after this one
 753          * @private
 754          * @deprecated use $this->mStripState->unstrip()
 755          */
 756         function unstrip( $text, $state ) {
 757                 return $state->unstripGeneral( $text );
 758         }
 759
 760         /**
 761          * Always call this after unstrip() to preserve the order
 762          *
 763          * @private
 764          * @deprecated use $this->mStripState->unstrip()
 765          */
 766         function unstripNoWiki( $text, $state ) {
 767                 return $state->unstripNoWiki( $text );
 768         }
 769
 770         /**
 771          * @deprecated use $this->mStripState->unstripBoth()
 772          */
 773         function unstripForHTML( $text ) {
 774                 return $this->mStripState->unstripBoth( $text );
 775         }
 776
 777         /**
 778          * Add an item to the strip state
 779          * Returns the unique tag which must be inserted into the stripped text
 780          * The tag will be replaced with the original text in unstrip()
 781          *
 782          * @private
 783          */
 784         function insertStripItem( $text ) {
 785                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 786                 $this->mMarkerIndex++;
 787                 $this->mStripState->general->setPair( $rnd, $text );
 788                 return $rnd;
 789         }
 790
 791         /**
 792          * Interface with html tidy
 793          * @deprecated Use MWTidy::tidy()
 794          */
 795         public static function tidy( $text ) {
 796                 wfDeprecated( __METHOD__ );
 797                 return MWTidy::tidy( $text );
 798         }
 799
 800         /**
 801          * parse the wiki syntax used to render tables
 802          *
 803          * @private
 804          */
 805         function doTableStuff( $text ) {
 806                 wfProfileIn( __METHOD__ );
 807
 808                 $lines = StringUtils::explode( "\n", $text );
 809                 $out = '';
 810                 $td_history = array(); # Is currently a td tag open?
 811                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 812                 $tr_history = array(); # Is currently a tr tag open?
 813                 $tr_attributes = array(); # history of tr attributes
 814                 $has_opened_tr = array(); # Did this table open a <tr> element?
 815                 $indent_level = 0; # indent level of the table
 816
 817                 foreach ( $lines as $outLine ) {
 818                         $line = trim( $outLine );
 819
 820                         if ( $line === '' ) { # empty line, go to next line
 821                                 $out .= $outLine."\n";
 822                                 continue;
 823                         }
 824
 825                         $first_character = $line[0];
 826                         $matches = array();
 827
 828                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 829                                 # First check if we are starting a new table
 830                                 $indent_level = strlen( $matches[1] );
 831
 832                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 833                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 834
 835                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 836                                 array_push( $td_history , false );
 837                                 array_push( $last_tag_history , '' );
 838                                 array_push( $tr_history , false );
 839                                 array_push( $tr_attributes , '' );
 840                                 array_push( $has_opened_tr , false );
 841                         } elseif ( count( $td_history ) == 0 ) {
 842                                 # Don't do any of the following
 843                                 $out .= $outLine."\n";
 844                                 continue;
 845                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 846                                 # We are ending a table
 847                                 $line = '</table>' . substr( $line , 2 );
 848                                 $last_tag = array_pop( $last_tag_history );
 849
 850                                 if ( !array_pop( $has_opened_tr ) ) {
 851                                         $line = "<tr><td></td></tr>{$line}";
 852                                 }
 853
 854                                 if ( array_pop( $tr_history ) ) {
 855                                         $line = "</tr>{$line}";
 856                                 }
 857
 858                                 if ( array_pop( $td_history ) ) {
 859                                         $line = "</{$last_tag}>{$line}";
 860                                 }
 861                                 array_pop( $tr_attributes );
 862                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 863                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 864                                 # Now we have a table row
 865                                 $line = preg_replace( '#^\|-+#', '', $line );
 866
 867                                 # Whats after the tag is now only attributes
 868                                 $attributes = $this->mStripState->unstripBoth( $line );
 869                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 870                                 array_pop( $tr_attributes );
 871                                 array_push( $tr_attributes, $attributes );
 872
 873                                 $line = '';
 874                                 $last_tag = array_pop( $last_tag_history );
 875                                 array_pop( $has_opened_tr );
 876                                 array_push( $has_opened_tr , true );
 877
 878                                 if ( array_pop( $tr_history ) ) {
 879                                         $line = '</tr>';
 880                                 }
 881
 882                                 if ( array_pop( $td_history ) ) {
 883                                         $line = "</{$last_tag}>{$line}";
 884                                 }
 885
 886                                 $outLine = $line;
 887                                 array_push( $tr_history , false );
 888                                 array_push( $td_history , false );
 889                                 array_push( $last_tag_history , '' );
 890                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 891                                 # This might be cell elements, td, th or captions
 892                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 893                                         $first_character = '+';
 894                                         $line = substr( $line , 1 );
 895                                 }
 896
 897                                 $line = substr( $line , 1 );
 898
 899                                 if ( $first_character === '!' ) {
 900                                         $line = str_replace( '!!' , '||' , $line );
 901                                 }
 902
 903                                 # Split up multiple cells on the same line.
 904                                 # FIXME : This can result in improper nesting of tags processed
 905                                 # by earlier parser steps, but should avoid splitting up eg
 906                                 # attribute values containing literal "||".
 907                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 908
 909                                 $outLine = '';
 910
 911                                 # Loop through each table cell
 912                                 foreach ( $cells as $cell ) {
 913                                         $previous = '';
 914                                         if ( $first_character !== '+' ) {
 915                                                 $tr_after = array_pop( $tr_attributes );
 916                                                 if ( !array_pop( $tr_history ) ) {
 917                                                         $previous = "<tr{$tr_after}>\n";
 918                                                 }
 919                                                 array_push( $tr_history , true );
 920                                                 array_push( $tr_attributes , '' );
 921                                                 array_pop( $has_opened_tr );
 922                                                 array_push( $has_opened_tr , true );
 923                                         }
 924
 925                                         $last_tag = array_pop( $last_tag_history );
 926
 927                                         if ( array_pop( $td_history ) ) {
 928                                                 $previous = "</{$last_tag}>\n{$previous}";
 929                                         }
 930
 931                                         if ( $first_character === '|' ) {
 932                                                 $last_tag = 'td';
 933                                         } elseif ( $first_character === '!' ) {
 934                                                 $last_tag = 'th';
 935                                         } elseif ( $first_character === '+' ) {
 936                                                 $last_tag = 'caption';
 937                                         } else {
 938                                                 $last_tag = '';
 939                                         }
 940
 941                                         array_push( $last_tag_history , $last_tag );
 942
 943                                         # A cell could contain both parameters and data
 944                                         $cell_data = explode( '|' , $cell , 2 );
 945
 946                                         # Bug 553: Note that a '|' inside an invalid link should not
 947                                         # be mistaken as delimiting cell parameters
 948                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 949                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 950                                         } elseif ( count( $cell_data ) == 1 ) {
 951                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 952                                         } else {
 953                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 954                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 955                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 956                                         }
 957
 958                                         $outLine .= $cell;
 959                                         array_push( $td_history , true );
 960                                 }
 961                         }
 962                         $out .= $outLine . "\n";
 963                 }
 964
 965                 # Closing open td, tr && table
 966                 while ( count( $td_history ) > 0 ) {
 967                         if ( array_pop( $td_history ) ) {
 968                                 $out .= "</td>\n";
 969                         }
 970                         if ( array_pop( $tr_history ) ) {
 971                                 $out .= "</tr>\n";
 972                         }
 973                         if ( !array_pop( $has_opened_tr ) ) {
 974                                 $out .= "<tr><td></td></tr>\n" ;
 975                         }
 976
 977                         $out .= "</table>\n";
 978                 }
 979
 980                 # Remove trailing line-ending (b/c)
 981                 if ( substr( $out, -1 ) === "\n" ) {
 982                         $out = substr( $out, 0, -1 );
 983                 }
 984
 985                 # special case: don't return empty table
 986                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 987                         $out = '';
 988                 }
 989
 990                 wfProfileOut( __METHOD__ );
 991
 992                 return $out;
 993         }
 994
 995         /**
 996          * Helper function for parse() that transforms wiki markup into
 997          * HTML. Only called for $mOutputType == self::OT_HTML.
 998          *
 999          * @private
1000          */
1001         function internalParse( $text, $isMain = true, $frame=false ) {
1002                 wfProfileIn( __METHOD__ );
1003
1004                 $origText = $text;
1005
1006                 # Hook to suspend the parser in this state
1007                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1008                         wfProfileOut( __METHOD__ );
1009                         return $text ;
1010                 }
1011
1012                 # if $frame is provided, then use $frame for replacing any variables
1013                 if ( $frame ) {
1014                         # use frame depth to infer how include/noinclude tags should be handled
1015                         # depth=0 means this is the top-level document; otherwise it's an included document
1016                         if ( !$frame->depth ) {
1017                                 $flag = 0;
1018                         } else {
1019                                 $flag = Parser::PTD_FOR_INCLUSION;
1020                         }
1021                         $dom = $this->preprocessToDom( $text, $flag );
1022                         $text = $frame->expand( $dom );
1023                 } else {
1024                         # if $frame is not provided, then use old-style replaceVariables
1025                         $text = $this->replaceVariables( $text );
1026                 }
1027
1028                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1029                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1030
1031                 # Tables need to come after variable replacement for things to work
1032                 # properly; putting them before other transformations should keep
1033                 # exciting things like link expansions from showing up in surprising
1034                 # places.
1035                 $text = $this->doTableStuff( $text );
1036
1037                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1038
1039                 $text = $this->doDoubleUnderscore( $text );
1040
1041                 $text = $this->doHeadings( $text );
1042                 if ( $this->mOptions->getUseDynamicDates() ) {
1043                         $df = DateFormatter::getInstance();
1044                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1045                 }
1046                 $text = $this->replaceInternalLinks( $text );
1047                 $text = $this->doAllQuotes( $text );
1048                 $text = $this->replaceExternalLinks( $text );
1049
1050                 # replaceInternalLinks may sometimes leave behind
1051                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1052                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1053
1054                 $text = $this->doMagicLinks( $text );
1055                 $text = $this->formatHeadings( $text, $origText, $isMain );
1056
1057                 wfProfileOut( __METHOD__ );
1058                 return $text;
1059         }
1060
1061         /**
1062          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1063          * magic external links.
1064          *
1065          * DML
1066          * @private
1067          */
1068         function doMagicLinks( $text ) {
1069                 wfProfileIn( __METHOD__ );
1070                 $prots = $this->mUrlProtocols;
1071                 $urlChar = self::EXT_LINK_URL_CLASS;
1072                 $text = preg_replace_callback(
1073                         '!(?:                           # Start cases
1074                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1075                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1076                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1077                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1078                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1079                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1080                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1081                                         [0-9Xx]                 # check digit
1082                                         \b)
1083                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1084                 wfProfileOut( __METHOD__ );
1085                 return $text;
1086         }
1087
1088         function magicLinkCallback( $m ) {
1089                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1090                         # Skip anchor
1091                         return $m[0];
1092                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1093                         # Skip HTML element
1094                         return $m[0];
1095                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1096                         # Free external link
1097                         return $this->makeFreeExternalLink( $m[0] );
1098                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1099                         # RFC or PMID
1100                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1101                                 $keyword = 'RFC';
1102                                 $urlmsg = 'rfcurl';
1103                                 $CssClass = 'mw-magiclink-rfc';
1104                                 $id = $m[4];
1105                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1106                                 $keyword = 'PMID';
1107                                 $urlmsg = 'pubmedurl';
1108                                 $CssClass = 'mw-magiclink-pmid';
1109                                 $id = $m[4];
1110                         } else {
1111                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1112                                         substr( $m[0], 0, 20 ) . '"' );
1113                         }
1114                         $url = wfMsgForContent( $urlmsg, $id );
1115                         $sk = $this->mOptions->getSkin( $this->mTitle );
1116                         return $sk->makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1117                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1118                         # ISBN
1119                         $isbn = $m[5];
1120                         $num = strtr( $isbn, array(
1121                                 '-' => '',
1122                                 ' ' => '',
1123                                 'x' => 'X',
1124                         ));
1125                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1126                         return'<a href="' .
1127                                 $titleObj->escapeLocalUrl() .
1128                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1129                 } else {
1130                         return $m[0];
1131                 }
1132         }
1133
1134         /**
1135          * Make a free external link, given a user-supplied URL
1136          * @return HTML
1137          * @private
1138          */
1139         function makeFreeExternalLink( $url ) {
1140                 global $wgContLang;
1141                 wfProfileIn( __METHOD__ );
1142
1143                 $sk = $this->mOptions->getSkin( $this->mTitle );
1144                 $trail = '';
1145
1146                 # The characters '<' and '>' (which were escaped by
1147                 # removeHTMLtags()) should not be included in
1148                 # URLs, per RFC 2396.
1149                 $m2 = array();
1150                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1151                         $trail = substr( $url, $m2[0][1] ) . $trail;
1152                         $url = substr( $url, 0, $m2[0][1] );
1153                 }
1154
1155                 # Move trailing punctuation to $trail
1156                 $sep = ',;\.:!?';
1157                 # If there is no left bracket, then consider right brackets fair game too
1158                 if ( strpos( $url, '(' ) === false ) {
1159                         $sep .= ')';
1160                 }
1161
1162                 $numSepChars = strspn( strrev( $url ), $sep );
1163                 if ( $numSepChars ) {
1164                         $trail = substr( $url, -$numSepChars ) . $trail;
1165                         $url = substr( $url, 0, -$numSepChars );
1166                 }
1167
1168                 $url = Sanitizer::cleanUrl( $url );
1169
1170                 # Is this an external image?
1171                 $text = $this->maybeMakeExternalImage( $url );
1172                 if ( $text === false ) {
1173                         # Not an image, make a link
1174                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1175                                 $this->getExternalLinkAttribs( $url ) );
1176                         # Register it in the output object...
1177                         # Replace unnecessary URL escape codes with their equivalent characters
1178                         $pasteurized = self::replaceUnusualEscapes( $url );
1179                         $this->mOutput->addExternalLink( $pasteurized );
1180                 }
1181                 wfProfileOut( __METHOD__ );
1182                 return $text . $trail;
1183         }
1184
1185
1186         /**
1187          * Parse headers and return html
1188          *
1189          * @private
1190          */
1191         function doHeadings( $text ) {
1192                 wfProfileIn( __METHOD__ );
1193                 for ( $i = 6; $i >= 1; --$i ) {
1194                         $h = str_repeat( '=', $i );
1195                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1196                           "<h$i>\\1</h$i>", $text );
1197                 }
1198                 wfProfileOut( __METHOD__ );
1199                 return $text;
1200         }
1201
1202         /**
1203          * Replace single quotes with HTML markup
1204          * @private
1205          * @return string the altered text
1206          */
1207         function doAllQuotes( $text ) {
1208                 wfProfileIn( __METHOD__ );
1209                 $outtext = '';
1210                 $lines = StringUtils::explode( "\n", $text );
1211                 foreach ( $lines as $line ) {
1212                         $outtext .= $this->doQuotes( $line ) . "\n";
1213                 }
1214                 $outtext = substr( $outtext, 0,-1 );
1215                 wfProfileOut( __METHOD__ );
1216                 return $outtext;
1217         }
1218
1219         /**
1220          * Helper function for doAllQuotes()
1221          */
1222         public function doQuotes( $text ) {
1223                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1224                 if ( count( $arr ) == 1 ) {
1225                         return $text;
1226                 } else {
1227                         # First, do some preliminary work. This may shift some apostrophes from
1228                         # being mark-up to being text. It also counts the number of occurrences
1229                         # of bold and italics mark-ups.
1230                         $numbold = 0;
1231                         $numitalics = 0;
1232                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1233                                 if ( ( $i % 2 ) == 1 ) {
1234                                         # If there are ever four apostrophes, assume the first is supposed to
1235                                         # be text, and the remaining three constitute mark-up for bold text.
1236                                         if ( strlen( $arr[$i] ) == 4 ) {
1237                                                 $arr[$i-1] .= "'";
1238                                                 $arr[$i] = "'''";
1239                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1240                                                 # If there are more than 5 apostrophes in a row, assume they're all
1241                                                 # text except for the last 5.
1242                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1243                                                 $arr[$i] = "'''''";
1244                                         }
1245                                         # Count the number of occurrences of bold and italics mark-ups.
1246                                         # We are not counting sequences of five apostrophes.
1247                                         if ( strlen( $arr[$i] ) == 2 ) {
1248                                                 $numitalics++;
1249                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1250                                                 $numbold++;
1251                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1252                                                 $numitalics++;
1253                                                 $numbold++;
1254                                         }
1255                                 }
1256                         }
1257
1258                         # If there is an odd number of both bold and italics, it is likely
1259                         # that one of the bold ones was meant to be an apostrophe followed
1260                         # by italics. Which one we cannot know for certain, but it is more
1261                         # likely to be one that has a single-letter word before it.
1262                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1263                                 $i = 0;
1264                                 $firstsingleletterword = -1;
1265                                 $firstmultiletterword = -1;
1266                                 $firstspace = -1;
1267                                 foreach ( $arr as $r ) {
1268                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1269                                                 $x1 = substr( $arr[$i-1], -1 );
1270                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1271                                                 if ( $x1 === ' ' ) {
1272                                                         if ( $firstspace == -1 ) {
1273                                                                 $firstspace = $i;
1274                                                         }
1275                                                 } elseif ( $x2 === ' ') {
1276                                                         if ( $firstsingleletterword == -1 ) {
1277                                                                 $firstsingleletterword = $i;
1278                                                         }
1279                                                 } else {
1280                                                         if ( $firstmultiletterword == -1 ) {
1281                                                                 $firstmultiletterword = $i;
1282                                                         }
1283                                                 }
1284                                         }
1285                                         $i++;
1286                                 }
1287
1288                                 # If there is a single-letter word, use it!
1289                                 if ( $firstsingleletterword > -1 ) {
1290                                         $arr[$firstsingleletterword] = "''";
1291                                         $arr[$firstsingleletterword-1] .= "'";
1292                                 } elseif ( $firstmultiletterword > -1 ) {
1293                                         # If not, but there's a multi-letter word, use that one.
1294                                         $arr[$firstmultiletterword] = "''";
1295                                         $arr[$firstmultiletterword-1] .= "'";
1296                                 } elseif ( $firstspace > -1 ) {
1297                                         # ... otherwise use the first one that has neither.
1298                                         # (notice that it is possible for all three to be -1 if, for example,
1299                                         # there is only one pentuple-apostrophe in the line)
1300                                         $arr[$firstspace] = "''";
1301                                         $arr[$firstspace-1] .= "'";
1302                                 }
1303                         }
1304
1305                         # Now let's actually convert our apostrophic mush to HTML!
1306                         $output = '';
1307                         $buffer = '';
1308                         $state = '';
1309                         $i = 0;
1310                         foreach ( $arr as $r ) {
1311                                 if ( ( $i % 2 ) == 0 ) {
1312                                         if ( $state === 'both' ) {
1313                                                 $buffer .= $r;
1314                                         } else {
1315                                                 $output .= $r;
1316                                         }
1317                                 } else {
1318                                         if ( strlen( $r ) == 2 ) {
1319                                                 if ( $state === 'i' ) {
1320                                                         $output .= '</i>'; $state = '';
1321                                                 } elseif ( $state === 'bi' ) {
1322                                                         $output .= '</i>'; $state = 'b';
1323                                                 } elseif ( $state === 'ib' ) {
1324                                                         $output .= '</b></i><b>'; $state = 'b';
1325                                                 } elseif ( $state === 'both' ) {
1326                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1327                                                 } else { # $state can be 'b' or ''
1328                                                         $output .= '<i>'; $state .= 'i';
1329                                                 }
1330                                         } elseif ( strlen( $r ) == 3 ) {
1331                                                 if ( $state === 'b' ) {
1332                                                         $output .= '</b>'; $state = '';
1333                                                 } elseif ( $state === 'bi' ) {
1334                                                         $output .= '</i></b><i>'; $state = 'i';
1335                                                 } elseif ( $state === 'ib' ) {
1336                                                         $output .= '</b>'; $state = 'i';
1337                                                 } elseif ( $state === 'both' ) {
1338                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1339                                                 } else { # $state can be 'i' or ''
1340                                                         $output .= '<b>'; $state .= 'b';
1341                                                 }
1342                                         } elseif ( strlen( $r ) == 5 ) {
1343                                                 if ( $state === 'b' ) {
1344                                                         $output .= '</b><i>'; $state = 'i';
1345                                                 } elseif ( $state === 'i' ) {
1346                                                         $output .= '</i><b>'; $state = 'b';
1347                                                 } elseif ( $state === 'bi' ) {
1348                                                         $output .= '</i></b>'; $state = '';
1349                                                 } elseif ( $state === 'ib' ) {
1350                                                         $output .= '</b></i>'; $state = '';
1351                                                 } elseif ( $state === 'both' ) {
1352                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1353                                                 } else { # ($state == '')
1354                                                         $buffer = ''; $state = 'both';
1355                                                 }
1356                                         }
1357                                 }
1358                                 $i++;
1359                         }
1360                         # Now close all remaining tags.  Notice that the order is important.
1361                         if ( $state === 'b' || $state === 'ib' ) {
1362                                 $output .= '</b>';
1363                         }
1364                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1365                                 $output .= '</i>';
1366                         }
1367                         if ( $state === 'bi' ) {
1368                                 $output .= '</b>';
1369                         }
1370                         # There might be lonely ''''', so make sure we have a buffer
1371                         if ( $state === 'both' && $buffer ) {
1372                                 $output .= '<b><i>'.$buffer.'</i></b>';
1373                         }
1374                         return $output;
1375                 }
1376         }
1377
1378         /**
1379          * Replace external links (REL)
1380          *
1381          * Note: this is all very hackish and the order of execution matters a lot.
1382          * Make sure to run maintenance/parserTests.php if you change this code.
1383          *
1384          * @private
1385          */
1386         function replaceExternalLinks( $text ) {
1387                 global $wgContLang;
1388                 wfProfileIn( __METHOD__ );
1389
1390                 $sk = $this->mOptions->getSkin( $this->mTitle );
1391
1392                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1393                 $s = array_shift( $bits );
1394
1395                 $i = 0;
1396                 while ( $i<count( $bits ) ) {
1397                         $url = $bits[$i++];
1398                         $protocol = $bits[$i++];
1399                         $text = $bits[$i++];
1400                         $trail = $bits[$i++];
1401
1402                         # The characters '<' and '>' (which were escaped by
1403                         # removeHTMLtags()) should not be included in
1404                         # URLs, per RFC 2396.
1405                         $m2 = array();
1406                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1407                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1408                                 $url = substr( $url, 0, $m2[0][1] );
1409                         }
1410
1411                         # If the link text is an image URL, replace it with an <img> tag
1412                         # This happened by accident in the original parser, but some people used it extensively
1413                         $img = $this->maybeMakeExternalImage( $text );
1414                         if ( $img !== false ) {
1415                                 $text = $img;
1416                         }
1417
1418                         $dtrail = '';
1419
1420                         # Set linktype for CSS - if URL==text, link is essentially free
1421                         $linktype = ( $text === $url ) ? 'free' : 'text';
1422
1423                         # No link text, e.g. [http://domain.tld/some.link]
1424                         if ( $text == '' ) {
1425                                 # Autonumber if allowed. See bug #5918
1426                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1427                                         $langObj = $this->getFunctionLang();
1428                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1429                                         $linktype = 'autonumber';
1430                                 } else {
1431                                         # Otherwise just use the URL
1432                                         $text = htmlspecialchars( $url );
1433                                         $linktype = 'free';
1434                                 }
1435                         } else {
1436                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1437                                 # Check for trail
1438                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1439                         }
1440
1441                         $text = $wgContLang->markNoConversion( $text );
1442
1443                         $url = Sanitizer::cleanUrl( $url );
1444
1445                         # Use the encoded URL
1446                         # This means that users can paste URLs directly into the text
1447                         # Funny characters like ö aren't valid in URLs anyway
1448                         # This was changed in August 2004
1449                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1450                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1451
1452                         # Register link in the output object.
1453                         # Replace unnecessary URL escape codes with the referenced character
1454                         # This prevents spammers from hiding links from the filters
1455                         $pasteurized = self::replaceUnusualEscapes( $url );
1456                         $this->mOutput->addExternalLink( $pasteurized );
1457                 }
1458
1459                 wfProfileOut( __METHOD__ );
1460                 return $s;
1461         }
1462
1463         /**
1464          * Get an associative array of additional HTML attributes appropriate for a
1465          * particular external link.  This currently may include rel => nofollow
1466          * (depending on configuration, namespace, and the URL's domain) and/or a
1467          * target attribute (depending on configuration).
1468          *
1469          * @param $url String: optional URL, to extract the domain from for rel =>
1470          *   nofollow if appropriate
1471          * @return Array: associative array of HTML attributes
1472          */
1473         function getExternalLinkAttribs( $url = false ) {
1474                 $attribs = array();
1475                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1476                 $ns = $this->mTitle->getNamespace();
1477                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1478                         $attribs['rel'] = 'nofollow';
1479
1480                         global $wgNoFollowDomainExceptions;
1481                         if ( $wgNoFollowDomainExceptions ) {
1482                                 $bits = wfParseUrl( $url );
1483                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1484                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1485                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1486                                                         unset( $attribs['rel'] );
1487                                                         break;
1488                                                 }
1489                                         }
1490                                 }
1491                         }
1492                 }
1493                 if ( $this->mOptions->getExternalLinkTarget() ) {
1494                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1495                 }
1496                 return $attribs;
1497         }
1498
1499
1500         /**
1501          * Replace unusual URL escape codes with their equivalent characters
1502          *
1503          * @param $url String
1504          * @return String
1505          *
1506          * @todo  This can merge genuinely required bits in the path or query string,
1507          *        breaking legit URLs. A proper fix would treat the various parts of
1508          *        the URL differently; as a workaround, just use the output for
1509          *        statistical records, not for actual linking/output.
1510          */
1511         static function replaceUnusualEscapes( $url ) {
1512                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1513                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1514         }
1515
1516         /**
1517          * Callback function used in replaceUnusualEscapes().
1518          * Replaces unusual URL escape codes with their equivalent character
1519          */
1520         private static function replaceUnusualEscapesCallback( $matches ) {
1521                 $char = urldecode( $matches[0] );
1522                 $ord = ord( $char );
1523                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1524                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1525                         # No, shouldn't be escaped
1526                         return $char;
1527                 } else {
1528                         # Yes, leave it escaped
1529                         return $matches[0];
1530                 }
1531         }
1532
1533         /**
1534          * make an image if it's allowed, either through the global
1535          * option, through the exception, or through the on-wiki whitelist
1536          * @private
1537          */
1538         function maybeMakeExternalImage( $url ) {
1539                 $sk = $this->mOptions->getSkin( $this->mTitle );
1540                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1541                 $imagesexception = !empty( $imagesfrom );
1542                 $text = false;
1543                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1544                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1545                         $imagematch = false;
1546                         foreach ( $imagesfrom as $match ) {
1547                                 if ( strpos( $url, $match ) === 0 ) {
1548                                         $imagematch = true;
1549                                         break;
1550                                 }
1551                         }
1552                 } elseif ( $imagesexception ) {
1553                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1554                 } else {
1555                         $imagematch = false;
1556                 }
1557                 if ( $this->mOptions->getAllowExternalImages()
1558                          || ( $imagesexception && $imagematch ) ) {
1559                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1560                                 # Image found
1561                                 $text = $sk->makeExternalImage( $url );
1562                         }
1563                 }
1564                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1565                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1566                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1567                         foreach ( $whitelist as $entry ) {
1568                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1569                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1570                                         continue;
1571                                 }
1572                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1573                                         # Image matches a whitelist entry
1574                                         $text = $sk->makeExternalImage( $url );
1575                                         break;
1576                                 }
1577                         }
1578                 }
1579                 return $text;
1580         }
1581
1582         /**
1583          * Process [[ ]] wikilinks
1584          * @return String: processed text
1585          *
1586          * @private
1587          */
1588         function replaceInternalLinks( $s ) {
1589                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1590                 return $s;
1591         }
1592
1593         /**
1594          * Process [[ ]] wikilinks (RIL)
1595          * @return LinkHolderArray
1596          *
1597          * @private
1598          */
1599         function replaceInternalLinks2( &$s ) {
1600                 global $wgContLang;
1601
1602                 wfProfileIn( __METHOD__ );
1603
1604                 wfProfileIn( __METHOD__.'-setup' );
1605                 static $tc = FALSE, $e1, $e1_img;
1606                 # the % is needed to support urlencoded titles as well
1607                 if ( !$tc ) {
1608                         $tc = Title::legalChars() . '#%';
1609                         # Match a link having the form [[namespace:link|alternate]]trail
1610                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1611                         # Match cases where there is no "]]", which might still be images
1612                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1613                 }
1614
1615                 $sk = $this->mOptions->getSkin( $this->mTitle );
1616                 $holders = new LinkHolderArray( $this );
1617
1618                 # split the entire text string on occurences of [[
1619                 $a = StringUtils::explode( '[[', ' ' . $s );
1620                 # get the first element (all text up to first [[), and remove the space we added
1621                 $s = $a->current();
1622                 $a->next();
1623                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1624                 $s = substr( $s, 1 );
1625
1626                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1627                 $e2 = null;
1628                 if ( $useLinkPrefixExtension ) {
1629                         # Match the end of a line for a word that's not followed by whitespace,
1630                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1631                         $e2 = wfMsgForContent( 'linkprefix' );
1632                 }
1633
1634                 if ( is_null( $this->mTitle ) ) {
1635                         wfProfileOut( __METHOD__.'-setup' );
1636                         wfProfileOut( __METHOD__ );
1637                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1638                 }
1639                 $nottalk = !$this->mTitle->isTalkPage();
1640
1641                 if ( $useLinkPrefixExtension ) {
1642                         $m = array();
1643                         if ( preg_match( $e2, $s, $m ) ) {
1644                                 $first_prefix = $m[2];
1645                         } else {
1646                                 $first_prefix = false;
1647                         }
1648                 } else {
1649                         $prefix = '';
1650                 }
1651
1652                 if ( $wgContLang->hasVariants() ) {
1653                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1654                 } else {
1655                         $selflink = array( $this->mTitle->getPrefixedText() );
1656                 }
1657                 $useSubpages = $this->areSubpagesAllowed();
1658                 wfProfileOut( __METHOD__.'-setup' );
1659
1660                 # Loop for each link
1661                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1662                         # Check for excessive memory usage
1663                         if ( $holders->isBig() ) {
1664                                 # Too big
1665                                 # Do the existence check, replace the link holders and clear the array
1666                                 $holders->replace( $s );
1667                                 $holders->clear();
1668                         }
1669
1670                         if ( $useLinkPrefixExtension ) {
1671                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1672                                 if ( preg_match( $e2, $s, $m ) ) {
1673                                         $prefix = $m[2];
1674                                         $s = $m[1];
1675                                 } else {
1676                                         $prefix='';
1677                                 }
1678                                 # first link
1679                                 if ( $first_prefix ) {
1680                                         $prefix = $first_prefix;
1681                                         $first_prefix = false;
1682                                 }
1683                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1684                         }
1685
1686                         $might_be_img = false;
1687
1688                         wfProfileIn( __METHOD__."-e1" );
1689                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1690                                 $text = $m[2];
1691                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1692                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1693                                 # the real problem is with the $e1 regex
1694                                 # See bug 1300.
1695                                 #
1696                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1697                                 # and no image is in sight. See bug 2095.
1698                                 #
1699                                 if ( $text !== '' &&
1700                                         substr( $m[3], 0, 1 ) === ']' &&
1701                                         strpos( $text, '[' ) !== false
1702                                 )
1703                                 {
1704                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1705                                         $m[3] = substr( $m[3], 1 );
1706                                 }
1707                                 # fix up urlencoded title texts
1708                                 if ( strpos( $m[1], '%' ) !== false ) {
1709                                         # Should anchors '#' also be rejected?
1710                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1711                                 }
1712                                 $trail = $m[3];
1713                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1714                                 $might_be_img = true;
1715                                 $text = $m[2];
1716                                 if ( strpos( $m[1], '%' ) !== false ) {
1717                                         $m[1] = rawurldecode( $m[1] );
1718                                 }
1719                                 $trail = "";
1720                         } else { # Invalid form; output directly
1721                                 $s .= $prefix . '[[' . $line ;
1722                                 wfProfileOut( __METHOD__."-e1" );
1723                                 continue;
1724                         }
1725                         wfProfileOut( __METHOD__."-e1" );
1726                         wfProfileIn( __METHOD__."-misc" );
1727
1728                         # Don't allow internal links to pages containing
1729                         # PROTO: where PROTO is a valid URL protocol; these
1730                         # should be external links.
1731                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1732                                 $s .= $prefix . '[[' . $line ;
1733                                 wfProfileOut( __METHOD__."-misc" );
1734                                 continue;
1735                         }
1736
1737                         # Make subpage if necessary
1738                         if ( $useSubpages ) {
1739                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1740                         } else {
1741                                 $link = $m[1];
1742                         }
1743
1744                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1745                         if ( !$noforce ) {
1746                                 # Strip off leading ':'
1747                                 $link = substr( $link, 1 );
1748                         }
1749
1750                         wfProfileOut( __METHOD__."-misc" );
1751                         wfProfileIn( __METHOD__."-title" );
1752                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1753                         if ( $nt === null ) {
1754                                 $s .= $prefix . '[[' . $line;
1755                                 wfProfileOut( __METHOD__."-title" );
1756                                 continue;
1757                         }
1758
1759                         $ns = $nt->getNamespace();
1760                         $iw = $nt->getInterWiki();
1761                         wfProfileOut( __METHOD__."-title" );
1762
1763                         if ( $might_be_img ) { # if this is actually an invalid link
1764                                 wfProfileIn( __METHOD__."-might_be_img" );
1765                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1766                                         $found = false;
1767                                         while ( true ) {
1768                                                 # look at the next 'line' to see if we can close it there
1769                                                 $a->next();
1770                                                 $next_line = $a->current();
1771                                                 if ( $next_line === false || $next_line === null ) {
1772                                                         break;
1773                                                 }
1774                                                 $m = explode( ']]', $next_line, 3 );
1775                                                 if ( count( $m ) == 3 ) {
1776                                                         # the first ]] closes the inner link, the second the image
1777                                                         $found = true;
1778                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1779                                                         $trail = $m[2];
1780                                                         break;
1781                                                 } elseif ( count( $m ) == 2 ) {
1782                                                         # if there's exactly one ]] that's fine, we'll keep looking
1783                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1784                                                 } else {
1785                                                         # if $next_line is invalid too, we need look no further
1786                                                         $text .= '[[' . $next_line;
1787                                                         break;
1788                                                 }
1789                                         }
1790                                         if ( !$found ) {
1791                                                 # we couldn't find the end of this imageLink, so output it raw
1792                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1793                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1794                                                 $s .= "{$prefix}[[$link|$text";
1795                                                 # note: no $trail, because without an end, there *is* no trail
1796                                                 wfProfileOut( __METHOD__."-might_be_img" );
1797                                                 continue;
1798                                         }
1799                                 } else { # it's not an image, so output it raw
1800                                         $s .= "{$prefix}[[$link|$text";
1801                                         # note: no $trail, because without an end, there *is* no trail
1802                                         wfProfileOut( __METHOD__."-might_be_img" );
1803                                         continue;
1804                                 }
1805                                 wfProfileOut( __METHOD__."-might_be_img" );
1806                         }
1807
1808                         $wasblank = ( $text  == '' );
1809                         if ( $wasblank ) {
1810                                 $text = $link;
1811                         } else {
1812                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1813                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1814                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1815                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1816                                 $text = $this->doQuotes( $text );
1817                         }
1818
1819                         # Link not escaped by : , create the various objects
1820                         if ( $noforce ) {
1821
1822                                 # Interwikis
1823                                 wfProfileIn( __METHOD__."-interwiki" );
1824                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1825                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1826                                         $s = rtrim( $s . $prefix );
1827                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1828                                         wfProfileOut( __METHOD__."-interwiki" );
1829                                         continue;
1830                                 }
1831                                 wfProfileOut( __METHOD__."-interwiki" );
1832
1833                                 if ( $ns == NS_FILE ) {
1834                                         wfProfileIn( __METHOD__."-image" );
1835                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1836                                                 if ( $wasblank ) {
1837                                                         # if no parameters were passed, $text
1838                                                         # becomes something like "File:Foo.png",
1839                                                         # which we don't want to pass on to the
1840                                                         # image generator
1841                                                         $text = '';
1842                                                 } else {
1843                                                         # recursively parse links inside the image caption
1844                                                         # actually, this will parse them in any other parameters, too,
1845                                                         # but it might be hard to fix that, and it doesn't matter ATM
1846                                                         $text = $this->replaceExternalLinks( $text );
1847                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1848                                                 }
1849                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1850                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1851                                         } else {
1852                                                 $s .= $prefix . $trail;
1853                                         }
1854                                         $this->mOutput->addImage( $nt->getDBkey() );
1855                                         wfProfileOut( __METHOD__."-image" );
1856                                         continue;
1857
1858                                 }
1859
1860                                 if ( $ns == NS_CATEGORY ) {
1861                                         wfProfileIn( __METHOD__."-category" );
1862                                         $s = rtrim( $s . "\n" ); # bug 87
1863
1864                                         if ( $wasblank ) {
1865                                                 $sortkey = $this->getDefaultSort();
1866                                         } else {
1867                                                 $sortkey = $text;
1868                                         }
1869                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1870                                         $sortkey = str_replace( "\n", '', $sortkey );
1871                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1872                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1873
1874                                         /**
1875                                          * Strip the whitespace Category links produce, see bug 87
1876                                          * @todo We might want to use trim($tmp, "\n") here.
1877                                          */
1878                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1879
1880                                         wfProfileOut( __METHOD__."-category" );
1881                                         continue;
1882                                 }
1883                         }
1884
1885                         # Self-link checking
1886                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1887                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1888                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1889                                         continue;
1890                                 }
1891                         }
1892
1893                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1894                         # FIXME: Should do batch file existence checks, see comment below
1895                         if ( $ns == NS_MEDIA ) {
1896                                 wfProfileIn( __METHOD__."-media" );
1897                                 # Give extensions a chance to select the file revision for us
1898                                 $skip = $time = false;
1899                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1900                                 if ( $skip ) {
1901                                         $link = $sk->link( $nt );
1902                                 } else {
1903                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1904                                 }
1905                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1906                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1907                                 $this->mOutput->addImage( $nt->getDBkey() );
1908                                 wfProfileOut( __METHOD__."-media" );
1909                                 continue;
1910                         }
1911
1912                         wfProfileIn( __METHOD__."-always_known" );
1913                         # Some titles, such as valid special pages or files in foreign repos, should
1914                         # be shown as bluelinks even though they're not included in the page table
1915                         #
1916                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1917                         # batch file existence checks for NS_FILE and NS_MEDIA
1918                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1919                                 $this->mOutput->addLink( $nt );
1920                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1921                         } else {
1922                                 # Links will be added to the output link list after checking
1923                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1924                         }
1925                         wfProfileOut( __METHOD__."-always_known" );
1926                 }
1927                 wfProfileOut( __METHOD__ );
1928                 return $holders;
1929         }
1930
1931         /**
1932          * Make a link placeholder. The text returned can be later resolved to a real link with
1933          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1934          * parsing of interwiki links, and secondly to allow all existence checks and
1935          * article length checks (for stub links) to be bundled into a single query.
1936          *
1937          * @deprecated
1938          */
1939         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1940                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1941         }
1942
1943         /**
1944          * Render a forced-blue link inline; protect against double expansion of
1945          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1946          * Since this little disaster has to split off the trail text to avoid
1947          * breaking URLs in the following text without breaking trails on the
1948          * wiki links, it's been made into a horrible function.
1949          *
1950          * @param $nt Title
1951          * @param $text String
1952          * @param $query String
1953          * @param $trail String
1954          * @param $prefix String
1955          * @return String: HTML-wikitext mix oh yuck
1956          */
1957         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1958                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1959                 $sk = $this->mOptions->getSkin( $this->mTitle );
1960                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1961                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1962                 return $this->armorLinks( $link ) . $trail;
1963         }
1964
1965         /**
1966          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1967          * going to go through further parsing steps before inline URL expansion.
1968          *
1969          * Not needed quite as much as it used to be since free links are a bit
1970          * more sensible these days. But bracketed links are still an issue.
1971          *
1972          * @param $text String: more-or-less HTML
1973          * @return String: less-or-more HTML with NOPARSE bits
1974          */
1975         function armorLinks( $text ) {
1976                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1977                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1978         }
1979
1980         /**
1981          * Return true if subpage links should be expanded on this page.
1982          * @return Boolean
1983          */
1984         function areSubpagesAllowed() {
1985                 # Some namespaces don't allow subpages
1986                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1987         }
1988
1989         /**
1990          * Handle link to subpage if necessary
1991          *
1992          * @param $target String: the source of the link
1993          * @param &$text String: the link text, modified as necessary
1994          * @return string the full name of the link
1995          * @private
1996          */
1997         function maybeDoSubpageLink( $target, &$text ) {
1998                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1999         }
2000
2001         /**#@+
2002          * Used by doBlockLevels()
2003          * @private
2004          */
2005         function closeParagraph() {
2006                 $result = '';
2007                 if ( $this->mLastSection != '' ) {
2008                         $result = '</' . $this->mLastSection  . ">\n";
2009                 }
2010                 $this->mInPre = false;
2011                 $this->mLastSection = '';
2012                 return $result;
2013         }
2014
2015         /**
2016          * getCommon() returns the length of the longest common substring
2017          * of both arguments, starting at the beginning of both.
2018          * @private
2019          */
2020         function getCommon( $st1, $st2 ) {
2021                 $fl = strlen( $st1 );
2022                 $shorter = strlen( $st2 );
2023                 if ( $fl < $shorter ) {
2024                         $shorter = $fl;
2025                 }
2026
2027                 for ( $i = 0; $i < $shorter; ++$i ) {
2028                         if ( $st1{$i} != $st2{$i} ) {
2029                                 break;
2030                         }
2031                 }
2032                 return $i;
2033         }
2034
2035         /**
2036          * These next three functions open, continue, and close the list
2037          * element appropriate to the prefix character passed into them.
2038          * @private
2039          */
2040         function openList( $char ) {
2041                 $result = $this->closeParagraph();
2042
2043                 if ( '*' === $char ) {
2044                         $result .= '<ul><li>';
2045                 } elseif ( '#' === $char ) {
2046                         $result .= '<ol><li>';
2047                 } elseif ( ':' === $char ) {
2048                         $result .= '<dl><dd>';
2049                 } elseif ( ';' === $char ) {
2050                         $result .= '<dl><dt>';
2051                         $this->mDTopen = true;
2052                 } else {
2053                         $result = '<!-- ERR 1 -->';
2054                 }
2055
2056                 return $result;
2057         }
2058
2059         /**
2060          * TODO: document
2061          * @param $char String
2062          * @private
2063          */
2064         function nextItem( $char ) {
2065                 if ( '*' === $char || '#' === $char ) {
2066                         return '</li><li>';
2067                 } elseif ( ':' === $char || ';' === $char ) {
2068                         $close = '</dd>';
2069                         if ( $this->mDTopen ) {
2070                                 $close = '</dt>';
2071                         }
2072                         if ( ';' === $char ) {
2073                                 $this->mDTopen = true;
2074                                 return $close . '<dt>';
2075                         } else {
2076                                 $this->mDTopen = false;
2077                                 return $close . '<dd>';
2078                         }
2079                 }
2080                 return '<!-- ERR 2 -->';
2081         }
2082
2083         /**
2084          * TODO: document
2085          * @param $char String
2086          * @private
2087          */
2088         function closeList( $char ) {
2089                 if ( '*' === $char ) {
2090                         $text = '</li></ul>';
2091                 } elseif ( '#' === $char ) {
2092                         $text = '</li></ol>';
2093                 } elseif ( ':' === $char ) {
2094                         if ( $this->mDTopen ) {
2095                                 $this->mDTopen = false;
2096                                 $text = '</dt></dl>';
2097                         } else {
2098                                 $text = '</dd></dl>';
2099                         }
2100                 } else {
2101                         return '<!-- ERR 3 -->';
2102                 }
2103                 return $text."\n";
2104         }
2105         /**#@-*/
2106
2107         /**
2108          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2109          *
2110          * @param $text String
2111          * @param $linestart Boolean: whether or not this is at the start of a line.
2112          * @private
2113          * @return string the lists rendered as HTML
2114          */
2115         function doBlockLevels( $text, $linestart ) {
2116                 wfProfileIn( __METHOD__ );
2117
2118                 # Parsing through the text line by line.  The main thing
2119                 # happening here is handling of block-level elements p, pre,
2120                 # and making lists from lines starting with * # : etc.
2121                 #
2122                 $textLines = StringUtils::explode( "\n", $text );
2123
2124                 $lastPrefix = $output = '';
2125                 $this->mDTopen = $inBlockElem = false;
2126                 $prefixLength = 0;
2127                 $paragraphStack = false;
2128
2129                 foreach ( $textLines as $oLine ) {
2130                         # Fix up $linestart
2131                         if ( !$linestart ) {
2132                                 $output .= $oLine;
2133                                 $linestart = true;
2134                                 continue;
2135                         }
2136                         # * = ul
2137                         # # = ol
2138                         # ; = dt
2139                         # : = dd
2140
2141                         $lastPrefixLength = strlen( $lastPrefix );
2142                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2143                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2144                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2145                         if ( !$this->mInPre ) {
2146                                 # Multiple prefixes may abut each other for nested lists.
2147                                 $prefixLength = strspn( $oLine, '*#:;' );
2148                                 $prefix = substr( $oLine, 0, $prefixLength );
2149
2150                                 # eh?
2151                                 # ; and : are both from definition-lists, so they're equivalent
2152                                 #  for the purposes of determining whether or not we need to open/close
2153                                 #  elements.
2154                                 $prefix2 = str_replace( ';', ':', $prefix );
2155                                 $t = substr( $oLine, $prefixLength );
2156                                 $this->mInPre = (bool)$preOpenMatch;
2157                         } else {
2158                                 # Don't interpret any other prefixes in preformatted text
2159                                 $prefixLength = 0;
2160                                 $prefix = $prefix2 = '';
2161                                 $t = $oLine;
2162                         }
2163
2164                         # List generation
2165                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2166                                 # Same as the last item, so no need to deal with nesting or opening stuff
2167                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2168                                 $paragraphStack = false;
2169
2170                                 if ( substr( $prefix, -1 ) === ';') {
2171                                         # The one nasty exception: definition lists work like this:
2172                                         # ; title : definition text
2173                                         # So we check for : in the remainder text to split up the
2174                                         # title and definition, without b0rking links.
2175                                         $term = $t2 = '';
2176                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2177                                                 $t = $t2;
2178                                                 $output .= $term . $this->nextItem( ':' );
2179                                         }
2180                                 }
2181                         } elseif ( $prefixLength || $lastPrefixLength ) {
2182                                 # We need to open or close prefixes, or both.
2183
2184                                 # Either open or close a level...
2185                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2186                                 $paragraphStack = false;
2187
2188                                 # Close all the prefixes which aren't shared.
2189                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2190                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2191                                         --$lastPrefixLength;
2192                                 }
2193
2194                                 # Continue the current prefix if appropriate.
2195                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2196                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2197                                 }
2198
2199                                 # Open prefixes where appropriate.
2200                                 while ( $prefixLength > $commonPrefixLength ) {
2201                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2202                                         $output .= $this->openList( $char );
2203
2204                                         if ( ';' === $char ) {
2205                                                 # FIXME: This is dupe of code above
2206                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2207                                                         $t = $t2;
2208                                                         $output .= $term . $this->nextItem( ':' );
2209                                                 }
2210                                         }
2211                                         ++$commonPrefixLength;
2212                                 }
2213                                 $lastPrefix = $prefix2;
2214                         }
2215
2216                         # If we have no prefixes, go to paragraph mode.
2217                         if ( 0 == $prefixLength ) {
2218                                 wfProfileIn( __METHOD__."-paragraph" );
2219                                 # No prefix (not in list)--go to paragraph mode
2220                                 # XXX: use a stack for nestable elements like span, table and div
2221                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2222                                 $closematch = preg_match(
2223                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2224                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2225                                 if ( $openmatch or $closematch ) {
2226                                         $paragraphStack = false;
2227                                         # TODO bug 5718: paragraph closed
2228                                         $output .= $this->closeParagraph();
2229                                         if ( $preOpenMatch and !$preCloseMatch ) {
2230                                                 $this->mInPre = true;
2231                                         }
2232                                         $inBlockElem = !$closematch;
2233                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2234                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2235                                                 # pre
2236                                                 if ( $this->mLastSection !== 'pre' ) {
2237                                                         $paragraphStack = false;
2238                                                         $output .= $this->closeParagraph().'<pre>';
2239                                                         $this->mLastSection = 'pre';
2240                                                 }
2241                                                 $t = substr( $t, 1 );
2242                                         } else {
2243                                                 # paragraph
2244                                                 if ( trim( $t ) === '' ) {
2245                                                         if ( $paragraphStack ) {
2246                                                                 $output .= $paragraphStack.'<br />';
2247                                                                 $paragraphStack = false;
2248                                                                 $this->mLastSection = 'p';
2249                                                         } else {
2250                                                                 if ( $this->mLastSection !== 'p' ) {
2251                                                                         $output .= $this->closeParagraph();
2252                                                                         $this->mLastSection = '';
2253                                                                         $paragraphStack = '<p>';
2254                                                                 } else {
2255                                                                         $paragraphStack = '</p><p>';
2256                                                                 }
2257                                                         }
2258                                                 } else {
2259                                                         if ( $paragraphStack ) {
2260                                                                 $output .= $paragraphStack;
2261                                                                 $paragraphStack = false;
2262                                                                 $this->mLastSection = 'p';
2263                                                         } elseif ( $this->mLastSection !== 'p' ) {
2264                                                                 $output .= $this->closeParagraph().'<p>';
2265                                                                 $this->mLastSection = 'p';
2266                                                         }
2267                                                 }
2268                                         }
2269                                 }
2270                                 wfProfileOut( __METHOD__."-paragraph" );
2271                         }
2272                         # somewhere above we forget to get out of pre block (bug 785)
2273                         if ( $preCloseMatch && $this->mInPre ) {
2274                                 $this->mInPre = false;
2275                         }
2276                         if ( $paragraphStack === false ) {
2277                                 $output .= $t."\n";
2278                         }
2279                 }
2280                 while ( $prefixLength ) {
2281                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2282                         --$prefixLength;
2283                 }
2284                 if ( $this->mLastSection != '' ) {
2285                         $output .= '</' . $this->mLastSection . '>';
2286                         $this->mLastSection = '';
2287                 }
2288
2289                 wfProfileOut( __METHOD__ );
2290                 return $output;
2291         }
2292
2293         /**
2294          * Split up a string on ':', ignoring any occurences inside tags
2295          * to prevent illegal overlapping.
2296          *
2297          * @param $str String: the string to split
2298          * @param &$before String: set to everything before the ':'
2299          * @param &$after String: set to everything after the ':'
2300          * return String: the position of the ':', or false if none found
2301          */
2302         function findColonNoLinks( $str, &$before, &$after ) {
2303                 wfProfileIn( __METHOD__ );
2304
2305                 $pos = strpos( $str, ':' );
2306                 if ( $pos === false ) {
2307                         # Nothing to find!
2308                         wfProfileOut( __METHOD__ );
2309                         return false;
2310                 }
2311
2312                 $lt = strpos( $str, '<' );
2313                 if ( $lt === false || $lt > $pos ) {
2314                         # Easy; no tag nesting to worry about
2315                         $before = substr( $str, 0, $pos );
2316                         $after = substr( $str, $pos+1 );
2317                         wfProfileOut( __METHOD__ );
2318                         return $pos;
2319                 }
2320
2321                 # Ugly state machine to walk through avoiding tags.
2322                 $state = self::COLON_STATE_TEXT;
2323                 $stack = 0;
2324                 $len = strlen( $str );
2325                 for( $i = 0; $i < $len; $i++ ) {
2326                         $c = $str{$i};
2327
2328                         switch( $state ) {
2329                         # (Using the number is a performance hack for common cases)
2330                         case 0: # self::COLON_STATE_TEXT:
2331                                 switch( $c ) {
2332                                 case "<":
2333                                         # Could be either a <start> tag or an </end> tag
2334                                         $state = self::COLON_STATE_TAGSTART;
2335                                         break;
2336                                 case ":":
2337                                         if ( $stack == 0 ) {
2338                                                 # We found it!
2339                                                 $before = substr( $str, 0, $i );
2340                                                 $after = substr( $str, $i + 1 );
2341                                                 wfProfileOut( __METHOD__ );
2342                                                 return $i;
2343                                         }
2344                                         # Embedded in a tag; don't break it.
2345                                         break;
2346                                 default:
2347                                         # Skip ahead looking for something interesting
2348                                         $colon = strpos( $str, ':', $i );
2349                                         if ( $colon === false ) {
2350                                                 # Nothing else interesting
2351                                                 wfProfileOut( __METHOD__ );
2352                                                 return false;
2353                                         }
2354                                         $lt = strpos( $str, '<', $i );
2355                                         if ( $stack === 0 ) {
2356                                                 if ( $lt === false || $colon < $lt ) {
2357                                                         # We found it!
2358                                                         $before = substr( $str, 0, $colon );
2359                                                         $after = substr( $str, $colon + 1 );
2360                                                         wfProfileOut( __METHOD__ );
2361                                                         return $i;
2362                                                 }
2363                                         }
2364                                         if ( $lt === false ) {
2365                                                 # Nothing else interesting to find; abort!
2366                                                 # We're nested, but there's no close tags left. Abort!
2367                                                 break 2;
2368                                         }
2369                                         # Skip ahead to next tag start
2370                                         $i = $lt;
2371                                         $state = self::COLON_STATE_TAGSTART;
2372                                 }
2373                                 break;
2374                         case 1: # self::COLON_STATE_TAG:
2375                                 # In a <tag>
2376                                 switch( $c ) {
2377                                 case ">":
2378                                         $stack++;
2379                                         $state = self::COLON_STATE_TEXT;
2380                                         break;
2381                                 case "/":
2382                                         # Slash may be followed by >?
2383                                         $state = self::COLON_STATE_TAGSLASH;
2384                                         break;
2385                                 default:
2386                                         # ignore
2387                                 }
2388                                 break;
2389                         case 2: # self::COLON_STATE_TAGSTART:
2390                                 switch( $c ) {
2391                                 case "/":
2392                                         $state = self::COLON_STATE_CLOSETAG;
2393                                         break;
2394                                 case "!":
2395                                         $state = self::COLON_STATE_COMMENT;
2396                                         break;
2397                                 case ">":
2398                                         # Illegal early close? This shouldn't happen D:
2399                                         $state = self::COLON_STATE_TEXT;
2400                                         break;
2401                                 default:
2402                                         $state = self::COLON_STATE_TAG;
2403                                 }
2404                                 break;
2405                         case 3: # self::COLON_STATE_CLOSETAG:
2406                                 # In a </tag>
2407                                 if ( $c === ">" ) {
2408                                         $stack--;
2409                                         if ( $stack < 0 ) {
2410                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2411                                                 wfProfileOut( __METHOD__ );
2412                                                 return false;
2413                                         }
2414                                         $state = self::COLON_STATE_TEXT;
2415                                 }
2416                                 break;
2417                         case self::COLON_STATE_TAGSLASH:
2418                                 if ( $c === ">" ) {
2419                                         # Yes, a self-closed tag <blah/>
2420                                         $state = self::COLON_STATE_TEXT;
2421                                 } else {
2422                                         # Probably we're jumping the gun, and this is an attribute
2423                                         $state = self::COLON_STATE_TAG;
2424                                 }
2425                                 break;
2426                         case 5: # self::COLON_STATE_COMMENT:
2427                                 if ( $c === "-" ) {
2428                                         $state = self::COLON_STATE_COMMENTDASH;
2429                                 }
2430                                 break;
2431                         case self::COLON_STATE_COMMENTDASH:
2432                                 if ( $c === "-" ) {
2433                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2434                                 } else {
2435                                         $state = self::COLON_STATE_COMMENT;
2436                                 }
2437                                 break;
2438                         case self::COLON_STATE_COMMENTDASHDASH:
2439                                 if ( $c === ">" ) {
2440                                         $state = self::COLON_STATE_TEXT;
2441                                 } else {
2442                                         $state = self::COLON_STATE_COMMENT;
2443                                 }
2444                                 break;
2445                         default:
2446                                 throw new MWException( "State machine error in " . __METHOD__ );
2447                         }
2448                 }
2449                 if ( $stack > 0 ) {
2450                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2451                         wfProfileOut( __METHOD__ );
2452                         return false;
2453                 }
2454                 wfProfileOut( __METHOD__ );
2455                 return false;
2456         }
2457
2458         /**
2459          * Return value of a magic variable (like PAGENAME)
2460          *
2461          * @private
2462          */
2463         function getVariableValue( $index, $frame=false ) {
2464                 global $wgContLang, $wgSitename, $wgServer;
2465                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2466
2467                 /**
2468                  * Some of these require message or data lookups and can be
2469                  * expensive to check many times.
2470                  */
2471                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2472                         if ( isset( $this->mVarCache[$index] ) ) {
2473                                 return $this->mVarCache[$index];
2474                         }
2475                 }
2476
2477                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2478                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2479
2480                 # Use the time zone
2481                 global $wgLocaltimezone;
2482                 if ( isset( $wgLocaltimezone ) ) {
2483                         $oldtz = date_default_timezone_get();
2484                         date_default_timezone_set( $wgLocaltimezone );
2485                 }
2486
2487                 $localTimestamp = date( 'YmdHis', $ts );
2488                 $localMonth = date( 'm', $ts );
2489                 $localMonth1 = date( 'n', $ts );
2490                 $localMonthName = date( 'n', $ts );
2491                 $localDay = date( 'j', $ts );
2492                 $localDay2 = date( 'd', $ts );
2493                 $localDayOfWeek = date( 'w', $ts );
2494                 $localWeek = date( 'W', $ts );
2495                 $localYear = date( 'Y', $ts );
2496                 $localHour = date( 'H', $ts );
2497                 if ( isset( $wgLocaltimezone ) ) {
2498                         date_default_timezone_set( $oldtz );
2499                 }
2500
2501                 switch ( $index ) {
2502                         case 'currentmonth':
2503                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2504                                 break;
2505                         case 'currentmonth1':
2506                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2507                                 break;
2508                         case 'currentmonthname':
2509                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2510                                 break;
2511                         case 'currentmonthnamegen':
2512                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2513                                 break;
2514                         case 'currentmonthabbrev':
2515                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2516                                 break;
2517                         case 'currentday':
2518                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2519                                 break;
2520                         case 'currentday2':
2521                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2522                                 break;
2523                         case 'localmonth':
2524                                 $value = $wgContLang->formatNum( $localMonth );
2525                                 break;
2526                         case 'localmonth1':
2527                                 $value = $wgContLang->formatNum( $localMonth1 );
2528                                 break;
2529                         case 'localmonthname':
2530                                 $value = $wgContLang->getMonthName( $localMonthName );
2531                                 break;
2532                         case 'localmonthnamegen':
2533                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2534                                 break;
2535                         case 'localmonthabbrev':
2536                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2537                                 break;
2538                         case 'localday':
2539                                 $value = $wgContLang->formatNum( $localDay );
2540                                 break;
2541                         case 'localday2':
2542                                 $value = $wgContLang->formatNum( $localDay2 );
2543                                 break;
2544                         case 'pagename':
2545                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2546                                 break;
2547                         case 'pagenamee':
2548                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2549                                 break;
2550                         case 'fullpagename':
2551                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2552                                 break;
2553                         case 'fullpagenamee':
2554                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2555                                 break;
2556                         case 'subpagename':
2557                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2558                                 break;
2559                         case 'subpagenamee':
2560                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2561                                 break;
2562                         case 'basepagename':
2563                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2564                                 break;
2565                         case 'basepagenamee':
2566                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2567                                 break;
2568                         case 'talkpagename':
2569                                 if ( $this->mTitle->canTalk() ) {
2570                                         $talkPage = $this->mTitle->getTalkPage();
2571                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2572                                 } else {
2573                                         $value = '';
2574                                 }
2575                                 break;
2576                         case 'talkpagenamee':
2577                                 if ( $this->mTitle->canTalk() ) {
2578                                         $talkPage = $this->mTitle->getTalkPage();
2579                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2580                                 } else {
2581                                         $value = '';
2582                                 }
2583                                 break;
2584                         case 'subjectpagename':
2585                                 $subjPage = $this->mTitle->getSubjectPage();
2586                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2587                                 break;
2588                         case 'subjectpagenamee':
2589                                 $subjPage = $this->mTitle->getSubjectPage();
2590                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2591                                 break;
2592                         case 'revisionid':
2593                                 # Let the edit saving system know we should parse the page
2594                                 # *after* a revision ID has been assigned.
2595                                 $this->mOutput->setFlag( 'vary-revision' );
2596                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2597                                 $value = $this->mRevisionId;
2598                                 break;
2599                         case 'revisionday':
2600                                 # Let the edit saving system know we should parse the page
2601                                 # *after* a revision ID has been assigned. This is for null edits.
2602                                 $this->mOutput->setFlag( 'vary-revision' );
2603                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2604                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2605                                 break;
2606                         case 'revisionday2':
2607                                 # Let the edit saving system know we should parse the page
2608                                 # *after* a revision ID has been assigned. This is for null edits.
2609                                 $this->mOutput->setFlag( 'vary-revision' );
2610                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2611                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2612                                 break;
2613                         case 'revisionmonth':
2614                                 # Let the edit saving system know we should parse the page
2615                                 # *after* a revision ID has been assigned. This is for null edits.
2616                                 $this->mOutput->setFlag( 'vary-revision' );
2617                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2618                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2619                                 break;
2620                         case 'revisionmonth1':
2621                                 # Let the edit saving system know we should parse the page
2622                                 # *after* a revision ID has been assigned. This is for null edits.
2623                                 $this->mOutput->setFlag( 'vary-revision' );
2624                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2625                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2626                                 break;
2627                         case 'revisionyear':
2628                                 # Let the edit saving system know we should parse the page
2629                                 # *after* a revision ID has been assigned. This is for null edits.
2630                                 $this->mOutput->setFlag( 'vary-revision' );
2631                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2632                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2633                                 break;
2634                         case 'revisiontimestamp':
2635                                 # Let the edit saving system know we should parse the page
2636                                 # *after* a revision ID has been assigned. This is for null edits.
2637                                 $this->mOutput->setFlag( 'vary-revision' );
2638                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2639                                 $value = $this->getRevisionTimestamp();
2640                                 break;
2641                         case 'revisionuser':
2642                                 # Let the edit saving system know we should parse the page
2643                                 # *after* a revision ID has been assigned. This is for null edits.
2644                                 $this->mOutput->setFlag( 'vary-revision' );
2645                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2646                                 $value = $this->getRevisionUser();
2647                                 break;
2648                         case 'namespace':
2649                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2650                                 break;
2651                         case 'namespacee':
2652                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2653                                 break;
2654                         case 'talkspace':
2655                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2656                                 break;
2657                         case 'talkspacee':
2658                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2659                                 break;
2660                         case 'subjectspace':
2661                                 $value = $this->mTitle->getSubjectNsText();
2662                                 break;
2663                         case 'subjectspacee':
2664                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2665                                 break;
2666                         case 'currentdayname':
2667                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2668                                 break;
2669                         case 'currentyear':
2670                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2671                                 break;
2672                         case 'currenttime':
2673                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2674                                 break;
2675                         case 'currenthour':
2676                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2677                                 break;
2678                         case 'currentweek':
2679                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2680                                 # int to remove the padding
2681                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2682                                 break;
2683                         case 'currentdow':
2684                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2685                                 break;
2686                         case 'localdayname':
2687                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2688                                 break;
2689                         case 'localyear':
2690                                 $value = $wgContLang->formatNum( $localYear, true );
2691                                 break;
2692                         case 'localtime':
2693                                 $value = $wgContLang->time( $localTimestamp, false, false );
2694                                 break;
2695                         case 'localhour':
2696                                 $value = $wgContLang->formatNum( $localHour, true );
2697                                 break;
2698                         case 'localweek':
2699                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2700                                 # int to remove the padding
2701                                 $value = $wgContLang->formatNum( (int)$localWeek );
2702                                 break;
2703                         case 'localdow':
2704                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2705                                 break;
2706                         case 'numberofarticles':
2707                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2708                                 break;
2709                         case 'numberoffiles':
2710                                 $value = $wgContLang->formatNum( SiteStats::images() );
2711                                 break;
2712                         case 'numberofusers':
2713                                 $value = $wgContLang->formatNum( SiteStats::users() );
2714                                 break;
2715                         case 'numberofactiveusers':
2716                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2717                                 break;
2718                         case 'numberofpages':
2719                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2720                                 break;
2721                         case 'numberofadmins':
2722                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2723                                 break;
2724                         case 'numberofedits':
2725                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2726                                 break;
2727                         case 'numberofviews':
2728                                 $value = $wgContLang->formatNum( SiteStats::views() );
2729                                 break;
2730                         case 'currenttimestamp':
2731                                 $value = wfTimestamp( TS_MW, $ts );
2732                                 break;
2733                         case 'localtimestamp':
2734                                 $value = $localTimestamp;
2735                                 break;
2736                         case 'currentversion':
2737                                 $value = SpecialVersion::getVersion();
2738                                 break;
2739                         case 'articlepath':
2740                                 return $wgArticlePath;
2741                         case 'sitename':
2742                                 return $wgSitename;
2743                         case 'server':
2744                                 return $wgServer;
2745                         case 'servername':
2746                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2747                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2748                                 wfRestoreWarnings();
2749                                 return $serverName ? $serverName : $wgServer;
2750                         case 'scriptpath':
2751                                 return $wgScriptPath;
2752                         case 'stylepath':
2753                                 return $wgStylePath;
2754                         case 'directionmark':
2755                                 return $wgContLang->getDirMark();
2756                         case 'contentlanguage':
2757                                 global $wgLanguageCode;
2758                                 return $wgLanguageCode;
2759                         default:
2760                                 $ret = null;
2761                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2762                                         return $ret;
2763                                 } else {
2764                                         return null;
2765                                 }
2766                 }
2767
2768                 if ( $index )
2769                         $this->mVarCache[$index] = $value;
2770
2771                 return $value;
2772         }
2773
2774         /**
2775          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2776          *
2777          * @private
2778          */
2779         function initialiseVariables() {
2780                 wfProfileIn( __METHOD__ );
2781                 $variableIDs = MagicWord::getVariableIDs();
2782                 $substIDs = MagicWord::getSubstIDs();
2783
2784                 $this->mVariables = new MagicWordArray( $variableIDs );
2785                 $this->mSubstWords = new MagicWordArray( $substIDs );
2786                 wfProfileOut( __METHOD__ );
2787         }
2788
2789         /**
2790          * Preprocess some wikitext and return the document tree.
2791          * This is the ghost of replace_variables().
2792          *
2793          * @param $text String: The text to parse
2794          * @param $flags Integer: bitwise combination of:
2795          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2796          *                                     included. Default is to assume a direct page view.
2797          *
2798          * The generated DOM tree must depend only on the input text and the flags.
2799          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2800          *
2801          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2802          * change in the DOM tree for a given text, must be passed through the section identifier
2803          * in the section edit link and thus back to extractSections().
2804          *
2805          * The output of this function is currently only cached in process memory, but a persistent
2806          * cache may be implemented at a later date which takes further advantage of these strict
2807          * dependency requirements.
2808          *
2809          * @private
2810          */
2811         function preprocessToDom( $text, $flags = 0 ) {
2812                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2813                 return $dom;
2814         }
2815
2816         /**
2817          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2818          */
2819         public static function splitWhitespace( $s ) {
2820                 $ltrimmed = ltrim( $s );
2821                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2822                 $trimmed = rtrim( $ltrimmed );
2823                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2824                 if ( $diff > 0 ) {
2825                         $w2 = substr( $ltrimmed, -$diff );
2826                 } else {
2827                         $w2 = '';
2828                 }
2829                 return array( $w1, $trimmed, $w2 );
2830         }
2831
2832         /**
2833          * Replace magic variables, templates, and template arguments
2834          * with the appropriate text. Templates are substituted recursively,
2835          * taking care to avoid infinite loops.
2836          *
2837          * Note that the substitution depends on value of $mOutputType:
2838          *  self::OT_WIKI: only {{subst:}} templates
2839          *  self::OT_PREPROCESS: templates but not extension tags
2840          *  self::OT_HTML: all templates and extension tags
2841          *
2842          * @param $text String: the text to transform
2843          * @param $frame PPFrame Object describing the arguments passed to the template.
2844          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2845          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2846          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2847          * @private
2848          */
2849         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2850                 # Is there any text? Also, Prevent too big inclusions!
2851                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2852                         return $text;
2853                 }
2854                 wfProfileIn( __METHOD__ );
2855
2856                 if ( $frame === false ) {
2857                         $frame = $this->getPreprocessor()->newFrame();
2858                 } elseif ( !( $frame instanceof PPFrame ) ) {
2859                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2860                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2861                 }
2862
2863                 $dom = $this->preprocessToDom( $text );
2864                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2865                 $text = $frame->expand( $dom, $flags );
2866
2867                 wfProfileOut( __METHOD__ );
2868                 return $text;
2869         }
2870
2871         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2872         static function createAssocArgs( $args ) {
2873                 $assocArgs = array();
2874                 $index = 1;
2875                 foreach ( $args as $arg ) {
2876                         $eqpos = strpos( $arg, '=' );
2877                         if ( $eqpos === false ) {
2878                                 $assocArgs[$index++] = $arg;
2879                         } else {
2880                                 $name = trim( substr( $arg, 0, $eqpos ) );
2881                                 $value = trim( substr( $arg, $eqpos+1 ) );
2882                                 if ( $value === false ) {
2883                                         $value = '';
2884                                 }
2885                                 if ( $name !== false ) {
2886                                         $assocArgs[$name] = $value;
2887                                 }
2888                         }
2889                 }
2890
2891                 return $assocArgs;
2892         }
2893
2894         /**
2895          * Warn the user when a parser limitation is reached
2896          * Will warn at most once the user per limitation type
2897          *
2898          * @param $limitationType String: should be one of:
2899          *   'expensive-parserfunction' (corresponding messages:
2900          *       'expensive-parserfunction-warning',
2901          *       'expensive-parserfunction-category')
2902          *   'post-expand-template-argument' (corresponding messages:
2903          *       'post-expand-template-argument-warning',
2904          *       'post-expand-template-argument-category')
2905          *   'post-expand-template-inclusion' (corresponding messages:
2906          *       'post-expand-template-inclusion-warning',
2907          *       'post-expand-template-inclusion-category')
2908          * @param $current Current value
2909          * @param $max Maximum allowed, when an explicit limit has been
2910          *       exceeded, provide the values (optional)
2911          */
2912         function limitationWarn( $limitationType, $current=null, $max=null) {
2913                 # does no harm if $current and $max are present but are unnecessary for the message
2914                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2915                 $this->mOutput->addWarning( $warning );
2916                 $this->addTrackingCategory( "$limitationType-category" );
2917         }
2918
2919         /**
2920          * Return the text of a template, after recursively
2921          * replacing any variables or templates within the template.
2922          *
2923          * @param $piece Array: the parts of the template
2924          *  $piece['title']: the title, i.e. the part before the |
2925          *  $piece['parts']: the parameter array
2926          *  $piece['lineStart']: whether the brace was at the start of a line
2927          * @param $frame PPFrame The current frame, contains template arguments
2928          * @return String: the text of the template
2929          * @private
2930          */
2931         function braceSubstitution( $piece, $frame ) {
2932                 global $wgContLang, $wgNonincludableNamespaces;
2933                 wfProfileIn( __METHOD__ );
2934                 wfProfileIn( __METHOD__.'-setup' );
2935
2936                 # Flags
2937                 $found = false;             # $text has been filled
2938                 $nowiki = false;            # wiki markup in $text should be escaped
2939                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2940                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2941                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2942                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2943
2944                 # Title object, where $text came from
2945                 $title = null;
2946
2947                 # $part1 is the bit before the first |, and must contain only title characters.
2948                 # Various prefixes will be stripped from it later.
2949                 $titleWithSpaces = $frame->expand( $piece['title'] );
2950                 $part1 = trim( $titleWithSpaces );
2951                 $titleText = false;
2952
2953                 # Original title text preserved for various purposes
2954                 $originalTitle = $part1;
2955
2956                 # $args is a list of argument nodes, starting from index 0, not including $part1
2957                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
2958                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2959                 wfProfileOut( __METHOD__.'-setup' );
2960
2961                 # SUBST
2962                 wfProfileIn( __METHOD__.'-modifiers' );
2963                 if ( !$found ) {
2964
2965                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2966
2967                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2968                         # Decide whether to expand template or keep wikitext as-is.
2969                         if ( $this->ot['wiki'] ) {
2970                                 if ( $substMatch === false ) {
2971                                         $literal = true;  # literal when in PST with no prefix
2972                                 } else {
2973                                         $literal = false; # expand when in PST with subst: or safesubst:
2974                                 }
2975                         } else {
2976                                 if ( $substMatch == 'subst' ) {
2977                                         $literal = true;  # literal when not in PST with plain subst:
2978                                 } else {
2979                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2980                                 }
2981                         }
2982                         if ( $literal ) {
2983                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2984                                 $isLocalObj = true;
2985                                 $found = true;
2986                         }
2987                 }
2988
2989                 # Variables
2990                 if ( !$found && $args->getLength() == 0 ) {
2991                         $id = $this->mVariables->matchStartToEnd( $part1 );
2992                         if ( $id !== false ) {
2993                                 $text = $this->getVariableValue( $id, $frame );
2994                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2995                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
2996                                 }
2997                                 $found = true;
2998                         }
2999                 }
3000
3001                 # MSG, MSGNW and RAW
3002                 if ( !$found ) {
3003                         # Check for MSGNW:
3004                         $mwMsgnw = MagicWord::get( 'msgnw' );
3005                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3006                                 $nowiki = true;
3007                         } else {
3008                                 # Remove obsolete MSG:
3009                                 $mwMsg = MagicWord::get( 'msg' );
3010                                 $mwMsg->matchStartAndRemove( $part1 );
3011                         }
3012
3013                         # Check for RAW:
3014                         $mwRaw = MagicWord::get( 'raw' );
3015                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3016                                 $forceRawInterwiki = true;
3017                         }
3018                 }
3019                 wfProfileOut( __METHOD__.'-modifiers' );
3020
3021                 # Parser functions
3022                 if ( !$found ) {
3023                         wfProfileIn( __METHOD__ . '-pfunc' );
3024
3025                         $colonPos = strpos( $part1, ':' );
3026                         if ( $colonPos !== false ) {
3027                                 # Case sensitive functions
3028                                 $function = substr( $part1, 0, $colonPos );
3029                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3030                                         $function = $this->mFunctionSynonyms[1][$function];
3031                                 } else {
3032                                         # Case insensitive functions
3033                                         $function = $wgContLang->lc( $function );
3034                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3035                                                 $function = $this->mFunctionSynonyms[0][$function];
3036                                         } else {
3037                                                 $function = false;
3038                                         }
3039                                 }
3040                                 if ( $function ) {
3041                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3042                                         $initialArgs = array( &$this );
3043                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3044                                         if ( $flags & SFH_OBJECT_ARGS ) {
3045                                                 # Add a frame parameter, and pass the arguments as an array
3046                                                 $allArgs = $initialArgs;
3047                                                 $allArgs[] = $frame;
3048                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3049                                                         $funcArgs[] = $args->item( $i );
3050                                                 }
3051                                                 $allArgs[] = $funcArgs;
3052                                         } else {
3053                                                 # Convert arguments to plain text
3054                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3055                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3056                                                 }
3057                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3058                                         }
3059
3060                                         # Workaround for PHP bug 35229 and similar
3061                                         if ( !is_callable( $callback ) ) {
3062                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3063                                                 wfProfileOut( __METHOD__ );
3064                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3065                                         }
3066                                         $result = call_user_func_array( $callback, $allArgs );
3067                                         $found = true;
3068                                         $noparse = true;
3069                                         $preprocessFlags = 0;
3070
3071                                         if ( is_array( $result ) ) {
3072                                                 if ( isset( $result[0] ) ) {
3073                                                         $text = $result[0];
3074                                                         unset( $result[0] );
3075                                                 }
3076
3077                                                 # Extract flags into the local scope
3078                                                 # This allows callers to set flags such as nowiki, found, etc.
3079                                                 extract( $result );
3080                                         } else {
3081                                                 $text = $result;
3082                                         }
3083                                         if ( !$noparse ) {
3084                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3085                                                 $isChildObj = true;
3086                                         }
3087                                 }
3088                         }
3089                         wfProfileOut( __METHOD__ . '-pfunc' );
3090                 }
3091
3092                 # Finish mangling title and then check for loops.
3093                 # Set $title to a Title object and $titleText to the PDBK
3094                 if ( !$found ) {
3095                         $ns = NS_TEMPLATE;
3096                         # Split the title into page and subpage
3097                         $subpage = '';
3098                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3099                         if ( $subpage !== '' ) {
3100                                 $ns = $this->mTitle->getNamespace();
3101                         }
3102                         $title = Title::newFromText( $part1, $ns );
3103                         if ( $title ) {
3104                                 $titleText = $title->getPrefixedText();
3105                                 # Check for language variants if the template is not found
3106                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3107                                         $wgContLang->findVariantLink( $part1, $title, true );
3108                                 }
3109                                 # Do recursion depth check
3110                                 $limit = $this->mOptions->getMaxTemplateDepth();
3111                                 if ( $frame->depth >= $limit ) {
3112                                         $found = true;
3113                                         $text = '<span class="error">'
3114                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3115                                                 . '</span>';
3116                                 }
3117                         }
3118                 }
3119
3120                 # Load from database
3121                 if ( !$found && $title ) {
3122                         wfProfileIn( __METHOD__ . '-loadtpl' );
3123                         if ( !$title->isExternal() ) {
3124                                 if ( $title->getNamespace() == NS_SPECIAL
3125                                         && $this->mOptions->getAllowSpecialInclusion()
3126                                         && $this->ot['html'] )
3127                                 {
3128                                         $text = SpecialPage::capturePath( $title );
3129                                         if ( is_string( $text ) ) {
3130                                                 $found = true;
3131                                                 $isHTML = true;
3132                                                 $this->disableCache();
3133                                         }
3134                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3135                                         $found = false; # access denied
3136                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3137                                 } else {
3138                                         list( $text, $title ) = $this->getTemplateDom( $title );
3139                                         if ( $text !== false ) {
3140                                                 $found = true;
3141                                                 $isChildObj = true;
3142                                         }
3143                                 }
3144
3145                                 # If the title is valid but undisplayable, make a link to it
3146                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3147                                         $text = "[[:$titleText]]";
3148                                         $found = true;
3149                                 }
3150                         } elseif ( $title->isTrans() ) {
3151                                 # Interwiki transclusion
3152                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3153                                         $text = $this->interwikiTransclude( $title, 'render' );
3154                                         $isHTML = true;
3155                                 } else {
3156                                         $text = $this->interwikiTransclude( $title, 'raw' );
3157                                         # Preprocess it like a template
3158                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3159                                         $isChildObj = true;
3160                                 }
3161                                 $found = true;
3162                         }
3163
3164                         # Do infinite loop check
3165                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3166                         if ( !$frame->loopCheck( $title ) ) {
3167                                 $found = true;
3168                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3169                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3170                         }
3171                         wfProfileOut( __METHOD__ . '-loadtpl' );
3172                 }
3173
3174                 # If we haven't found text to substitute by now, we're done
3175                 # Recover the source wikitext and return it
3176                 if ( !$found ) {
3177                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3178                         wfProfileOut( __METHOD__ );
3179                         return array( 'object' => $text );
3180                 }
3181
3182                 # Expand DOM-style return values in a child frame
3183                 if ( $isChildObj ) {
3184                         # Clean up argument array
3185                         $newFrame = $frame->newChild( $args, $title );
3186
3187                         if ( $nowiki ) {
3188                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3189                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3190                                 # Expansion is eligible for the empty-frame cache
3191                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3192                                         $text = $this->mTplExpandCache[$titleText];
3193                                 } else {
3194                                         $text = $newFrame->expand( $text );
3195                                         $this->mTplExpandCache[$titleText] = $text;
3196                                 }
3197                         } else {
3198                                 # Uncached expansion
3199                                 $text = $newFrame->expand( $text );
3200                         }
3201                 }
3202                 if ( $isLocalObj && $nowiki ) {
3203                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3204                         $isLocalObj = false;
3205                 }
3206
3207                 # Replace raw HTML by a placeholder
3208                 # Add a blank line preceding, to prevent it from mucking up
3209                 # immediately preceding headings
3210                 if ( $isHTML ) {
3211                         $text = "\n\n" . $this->insertStripItem( $text );
3212                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3213                         # Escape nowiki-style return values
3214                         $text = wfEscapeWikiText( $text );
3215                 } elseif ( is_string( $text )
3216                         && !$piece['lineStart']
3217                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3218                 {
3219                         # Bug 529: if the template begins with a table or block-level
3220                         # element, it should be treated as beginning a new line.
3221                         # This behaviour is somewhat controversial.
3222                         $text = "\n" . $text;
3223                 }
3224
3225                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3226                         # Error, oversize inclusion
3227                         if ( $titleText !== false ) {
3228                                 # Make a working, properly escaped link if possible (bug 23588)
3229                                 $text = "[[:$titleText]]";
3230                         } else {
3231                                 # This will probably not be a working link, but at least it may
3232                                 # provide some hint of where the problem is
3233                                 preg_replace( '/^:/', '', $originalTitle );
3234                                 $text = "[[:$originalTitle]]";
3235                         }
3236                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3237                         $this->limitationWarn( 'post-expand-template-inclusion' );
3238                 }
3239
3240                 if ( $isLocalObj ) {
3241                         $ret = array( 'object' => $text );
3242                 } else {
3243                         $ret = array( 'text' => $text );
3244                 }
3245
3246                 wfProfileOut( __METHOD__ );
3247                 return $ret;
3248         }
3249
3250         /**
3251          * Get the semi-parsed DOM representation of a template with a given title,
3252          * and its redirect destination title. Cached.
3253          */
3254         function getTemplateDom( $title ) {
3255                 $cacheTitle = $title;
3256                 $titleText = $title->getPrefixedDBkey();
3257
3258                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3259                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3260                         $title = Title::makeTitle( $ns, $dbk );
3261                         $titleText = $title->getPrefixedDBkey();
3262                 }
3263                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3264                         return array( $this->mTplDomCache[$titleText], $title );
3265                 }
3266
3267                 # Cache miss, go to the database
3268                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3269
3270                 if ( $text === false ) {
3271                         $this->mTplDomCache[$titleText] = false;
3272                         return array( false, $title );
3273                 }
3274
3275                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3276                 $this->mTplDomCache[ $titleText ] = $dom;
3277
3278                 if ( !$title->equals( $cacheTitle ) ) {
3279                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3280                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3281                 }
3282
3283                 return array( $dom, $title );
3284         }
3285
3286         /**
3287          * Fetch the unparsed text of a template and register a reference to it.
3288          */
3289         function fetchTemplateAndTitle( $title ) {
3290                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3291                 $stuff = call_user_func( $templateCb, $title, $this );
3292                 $text = $stuff['text'];
3293                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3294                 if ( isset( $stuff['deps'] ) ) {
3295                         foreach ( $stuff['deps'] as $dep ) {
3296                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3297                         }
3298                 }
3299                 return array( $text, $finalTitle );
3300         }
3301
3302         function fetchTemplate( $title ) {
3303                 $rv = $this->fetchTemplateAndTitle( $title );
3304                 return $rv[0];
3305         }
3306
3307         /**
3308          * Static function to get a template
3309          * Can be overridden via ParserOptions::setTemplateCallback().
3310          */
3311         static function statelessFetchTemplate( $title, $parser=false ) {
3312                 $text = $skip = false;
3313                 $finalTitle = $title;
3314                 $deps = array();
3315
3316                 # Loop to fetch the article, with up to 1 redirect
3317                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3318                         # Give extensions a chance to select the revision instead
3319                         $id = false; # Assume current
3320                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3321
3322                         if ( $skip ) {
3323                                 $text = false;
3324                                 $deps[] = array(
3325                                         'title' => $title,
3326                                         'page_id' => $title->getArticleID(),
3327                                         'rev_id' => null );
3328                                 break;
3329                         }
3330                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3331                         $rev_id = $rev ? $rev->getId() : 0;
3332                         # If there is no current revision, there is no page
3333                         if ( $id === false && !$rev ) {
3334                                 $linkCache = LinkCache::singleton();
3335                                 $linkCache->addBadLinkObj( $title );
3336                         }
3337
3338                         $deps[] = array(
3339                                 'title' => $title,
3340                                 'page_id' => $title->getArticleID(),
3341                                 'rev_id' => $rev_id );
3342
3343                         if ( $rev ) {
3344                                 $text = $rev->getText();
3345                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3346                                 global $wgContLang;
3347                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3348                                 if ( !$message->exists() ) {
3349                                         $text = false;
3350                                         break;
3351                                 }
3352                                 $text = $message->plain();
3353                         } else {
3354                                 break;
3355                         }
3356                         if ( $text === false ) {
3357                                 break;
3358                         }
3359                         # Redirect?
3360                         $finalTitle = $title;
3361                         $title = Title::newFromRedirect( $text );
3362                 }
3363                 return array(
3364                         'text' => $text,
3365                         'finalTitle' => $finalTitle,
3366                         'deps' => $deps );
3367         }
3368
3369         /**
3370          * Transclude an interwiki link.
3371          */
3372         function interwikiTransclude( $title, $action ) {
3373                 global $wgEnableScaryTranscluding;
3374
3375                 if ( !$wgEnableScaryTranscluding ) {
3376                         return wfMsgForContent('scarytranscludedisabled');
3377                 }
3378
3379                 $url = $title->getFullUrl( "action=$action" );
3380
3381                 if ( strlen( $url ) > 255 ) {
3382                         return wfMsgForContent( 'scarytranscludetoolong' );
3383                 }
3384                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3385         }
3386
3387         function fetchScaryTemplateMaybeFromCache( $url ) {
3388                 global $wgTranscludeCacheExpiry;
3389                 $dbr = wfGetDB( DB_SLAVE );
3390                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3391                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3392                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3393                 if ( $obj ) {
3394                         return $obj->tc_contents;
3395                 }
3396
3397                 $text = Http::get( $url );
3398                 if ( !$text ) {
3399                         return wfMsgForContent( 'scarytranscludefailed', $url );
3400                 }
3401
3402                 $dbw = wfGetDB( DB_MASTER );
3403                 $dbw->replace( 'transcache', array('tc_url'), array(
3404                         'tc_url' => $url,
3405                         'tc_time' => $dbw->timestamp( time() ),
3406                         'tc_contents' => $text)
3407                 );
3408                 return $text;
3409         }
3410
3411
3412         /**
3413          * Triple brace replacement -- used for template arguments
3414          * @private
3415          */
3416         function argSubstitution( $piece, $frame ) {
3417                 wfProfileIn( __METHOD__ );
3418
3419                 $error = false;
3420                 $parts = $piece['parts'];
3421                 $nameWithSpaces = $frame->expand( $piece['title'] );
3422                 $argName = trim( $nameWithSpaces );
3423                 $object = false;
3424                 $text = $frame->getArgument( $argName );
3425                 if (  $text === false && $parts->getLength() > 0
3426                   && (
3427                         $this->ot['html']
3428                         || $this->ot['pre']
3429                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3430                   )
3431                 ) {
3432                         # No match in frame, use the supplied default
3433                         $object = $parts->item( 0 )->getChildren();
3434                 }
3435                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3436                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3437                         $this->limitationWarn( 'post-expand-template-argument' );
3438                 }
3439
3440                 if ( $text === false && $object === false ) {
3441                         # No match anywhere
3442                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3443                 }
3444                 if ( $error !== false ) {
3445                         $text .= $error;
3446                 }
3447                 if ( $object !== false ) {
3448                         $ret = array( 'object' => $object );
3449                 } else {
3450                         $ret = array( 'text' => $text );
3451                 }
3452
3453                 wfProfileOut( __METHOD__ );
3454                 return $ret;
3455         }
3456
3457         /**
3458          * Return the text to be used for a given extension tag.
3459          * This is the ghost of strip().
3460          *
3461          * @param $params Associative array of parameters:
3462          *     name       PPNode for the tag name
3463          *     attr       PPNode for unparsed text where tag attributes are thought to be
3464          *     attributes Optional associative array of parsed attributes
3465          *     inner      Contents of extension element
3466          *     noClose    Original text did not have a close tag
3467          * @param $frame PPFrame
3468          */
3469         function extensionSubstitution( $params, $frame ) {
3470                 $name = $frame->expand( $params['name'] );
3471                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3472                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3473                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3474
3475                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3476                         ( $this->ot['html'] || $this->ot['pre'] );
3477                 if ( $isFunctionTag ) {
3478                         $markerType = 'none';
3479                 } else {
3480                         $markerType = 'general';
3481                 }
3482                 if ( $this->ot['html'] || $isFunctionTag ) {
3483                         $name = strtolower( $name );
3484                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3485                         if ( isset( $params['attributes'] ) ) {
3486                                 $attributes = $attributes + $params['attributes'];
3487                         }
3488
3489                         if ( isset( $this->mTagHooks[$name] ) ) {
3490                                 # Workaround for PHP bug 35229 and similar
3491                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3492                                         throw new MWException( "Tag hook for $name is not callable\n" );
3493                                 }
3494                                 $output = call_user_func_array( $this->mTagHooks[$name],
3495                                         array( $content, $attributes, $this, $frame ) );
3496                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3497                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3498                                 if ( !is_callable( $callback ) ) {
3499                                         throw new MWException( "Tag hook for $name is not callable\n" );
3500                                 }
3501
3502                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3503                         } else {
3504                                 $output = '<span class="error">Invalid tag extension name: ' .
3505                                         htmlspecialchars( $name ) . '</span>';
3506                         }
3507
3508                         if ( is_array( $output ) ) {
3509                                 # Extract flags to local scope (to override $markerType)
3510                                 $flags = $output;
3511                                 $output = $flags[0];
3512                                 unset( $flags[0] );
3513                                 extract( $flags );
3514                         }
3515                 } else {
3516                         if ( is_null( $attrText ) ) {
3517                                 $attrText = '';
3518                         }
3519                         if ( isset( $params['attributes'] ) ) {
3520                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3521                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3522                                                 htmlspecialchars( $attrValue ) . '"';
3523                                 }
3524                         }
3525                         if ( $content === null ) {
3526                                 $output = "<$name$attrText/>";
3527                         } else {
3528                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3529                                 $output = "<$name$attrText>$content$close";
3530                         }
3531                 }
3532
3533                 if ( $markerType === 'none' ) {
3534                         return $output;
3535                 } elseif ( $markerType === 'nowiki' ) {
3536                         $this->mStripState->nowiki->setPair( $marker, $output );
3537                 } elseif ( $markerType === 'general' ) {
3538                         $this->mStripState->general->setPair( $marker, $output );
3539                 } else {
3540                         throw new MWException( __METHOD__.': invalid marker type' );
3541                 }
3542                 return $marker;
3543         }
3544
3545         /**
3546          * Increment an include size counter
3547          *
3548          * @param $type String: the type of expansion
3549          * @param $size Integer: the size of the text
3550          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3551          */
3552         function incrementIncludeSize( $type, $size ) {
3553                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3554                         return false;
3555                 } else {
3556                         $this->mIncludeSizes[$type] += $size;
3557                         return true;
3558                 }
3559         }
3560
3561         /**
3562          * Increment the expensive function count
3563          *
3564          * @return Boolean: false if the limit has been exceeded
3565          */
3566         function incrementExpensiveFunctionCount() {
3567                 global $wgExpensiveParserFunctionLimit;
3568                 $this->mExpensiveFunctionCount++;
3569                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3570                         return true;
3571                 }
3572                 return false;
3573         }
3574
3575         /**
3576          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3577          * Fills $this->mDoubleUnderscores, returns the modified text
3578          */
3579         function doDoubleUnderscore( $text ) {
3580                 wfProfileIn( __METHOD__ );
3581
3582                 # The position of __TOC__ needs to be recorded
3583                 $mw = MagicWord::get( 'toc' );
3584                 if ( $mw->match( $text ) ) {
3585                         $this->mShowToc = true;
3586                         $this->mForceTocPosition = true;
3587
3588                         # Set a placeholder. At the end we'll fill it in with the TOC.
3589                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3590
3591                         # Only keep the first one.
3592                         $text = $mw->replace( '', $text );
3593                 }
3594
3595                 # Now match and remove the rest of them
3596                 $mwa = MagicWord::getDoubleUnderscoreArray();
3597                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3598
3599                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3600                         $this->mOutput->mNoGallery = true;
3601                 }
3602                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3603                         $this->mShowToc = false;
3604                 }
3605                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3606                         $this->addTrackingCategory( 'hidden-category-category' );
3607                 }
3608                 # (bug 8068) Allow control over whether robots index a page.
3609                 #
3610                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3611                 # is not desirable, the last one on the page should win.
3612                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3613                         $this->mOutput->setIndexPolicy( 'noindex' );
3614                         $this->addTrackingCategory( 'noindex-category' );
3615                 }
3616                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3617                         $this->mOutput->setIndexPolicy( 'index' );
3618                         $this->addTrackingCategory( 'index-category' );
3619                 }
3620
3621                 # Cache all double underscores in the database
3622                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3623                         $this->mOutput->setProperty( $key, '' );
3624                 }
3625
3626                 wfProfileOut( __METHOD__ );
3627                 return $text;
3628         }
3629
3630         /**
3631          * Add a tracking category, getting the title from a system message,
3632          * or print a debug message if the title is invalid.
3633          *
3634          * @param $msg String: message key
3635          * @return Boolean: whether the addition was successful
3636          */
3637         protected function addTrackingCategory( $msg ) {
3638                 $cat = wfMsgForContent( $msg );
3639
3640                 # Allow tracking categories to be disabled by setting them to "-"
3641                 if ( $cat === '-' ) {
3642                         return false;
3643                 }
3644
3645                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3646                 if ( $containerCategory ) {
3647                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3648                         return true;
3649                 } else {
3650                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3651                         return false;
3652                 }
3653         }
3654
3655         /**
3656          * This function accomplishes several tasks:
3657          * 1) Auto-number headings if that option is enabled
3658          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3659          * 3) Add a Table of contents on the top for users who have enabled the option
3660          * 4) Auto-anchor headings
3661          *
3662          * It loops through all headlines, collects the necessary data, then splits up the
3663          * string and re-inserts the newly formatted headlines.
3664          *
3665          * @param $text String
3666          * @param $origText String: original, untouched wikitext
3667          * @param $isMain Boolean
3668          * @private
3669          */
3670         function formatHeadings( $text, $origText, $isMain=true ) {
3671                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3672
3673                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3674
3675                 # Inhibit editsection links if requested in the page
3676                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3677                         $showEditLink = 0;
3678                 } else {
3679                         $showEditLink = $this->mOptions->getEditSection();
3680                 }
3681                 if ( $showEditLink ) {
3682                         $this->mOutput->setEditSectionTokens( true );
3683                 }
3684
3685                 # Get all headlines for numbering them and adding funky stuff like [edit]
3686                 # links - this is for later, but we need the number of headlines right now
3687                 $matches = array();
3688                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3689
3690                 # if there are fewer than 4 headlines in the article, do not show TOC
3691                 # unless it's been explicitly enabled.
3692                 $enoughToc = $this->mShowToc &&
3693                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3694
3695                 # Allow user to stipulate that a page should have a "new section"
3696                 # link added via __NEWSECTIONLINK__
3697                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3698                         $this->mOutput->setNewSection( true );
3699                 }
3700
3701                 # Allow user to remove the "new section"
3702                 # link via __NONEWSECTIONLINK__
3703                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3704                         $this->mOutput->hideNewSection( true );
3705                 }
3706
3707                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3708                 # override above conditions and always show TOC above first header
3709                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3710                         $this->mShowToc = true;
3711                         $enoughToc = true;
3712                 }
3713
3714                 # We need this to perform operations on the HTML
3715                 $sk = $this->mOptions->getSkin( $this->mTitle );
3716
3717                 # headline counter
3718                 $headlineCount = 0;
3719                 $numVisible = 0;
3720
3721                 # Ugh .. the TOC should have neat indentation levels which can be
3722                 # passed to the skin functions. These are determined here
3723                 $toc = '';
3724                 $full = '';
3725                 $head = array();
3726                 $sublevelCount = array();
3727                 $levelCount = array();
3728                 $level = 0;
3729                 $prevlevel = 0;
3730                 $toclevel = 0;
3731                 $prevtoclevel = 0;
3732                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3733                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3734                 $oldType = $this->mOutputType;
3735                 $this->setOutputType( self::OT_WIKI );
3736                 $frame = $this->getPreprocessor()->newFrame();
3737                 $root = $this->preprocessToDom( $origText );
3738                 $node = $root->getFirstChild();
3739                 $byteOffset = 0;
3740                 $tocraw = array();
3741                 $refers = array();
3742
3743                 foreach ( $matches[3] as $headline ) {
3744                         $isTemplate = false;
3745                         $titleText = false;
3746                         $sectionIndex = false;
3747                         $numbering = '';
3748                         $markerMatches = array();
3749                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3750                                 $serial = $markerMatches[1];
3751                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3752                                 $isTemplate = ( $titleText != $baseTitleText );
3753                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3754                         }
3755
3756                         if ( $toclevel ) {
3757                                 $prevlevel = $level;
3758                         }
3759                         $level = $matches[1][$headlineCount];
3760
3761                         if ( $level > $prevlevel ) {
3762                                 # Increase TOC level
3763                                 $toclevel++;
3764                                 $sublevelCount[$toclevel] = 0;
3765                                 if ( $toclevel<$wgMaxTocLevel ) {
3766                                         $prevtoclevel = $toclevel;
3767                                         $toc .= $sk->tocIndent();
3768                                         $numVisible++;
3769                                 }
3770                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3771                                 # Decrease TOC level, find level to jump to
3772
3773                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3774                                         if ( $levelCount[$i] == $level ) {
3775                                                 # Found last matching level
3776                                                 $toclevel = $i;
3777                                                 break;
3778                                         } elseif ( $levelCount[$i] < $level ) {
3779                                                 # Found first matching level below current level
3780                                                 $toclevel = $i + 1;
3781                                                 break;
3782                                         }
3783                                 }
3784                                 if ( $i == 0 ) {
3785                                         $toclevel = 1;
3786                                 }
3787                                 if ( $toclevel<$wgMaxTocLevel ) {
3788                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3789                                                 # Unindent only if the previous toc level was shown :p
3790                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3791                                                 $prevtoclevel = $toclevel;
3792                                         } else {
3793                                                 $toc .= $sk->tocLineEnd();
3794                                         }
3795                                 }
3796                         } else {
3797                                 # No change in level, end TOC line
3798                                 if ( $toclevel<$wgMaxTocLevel ) {
3799                                         $toc .= $sk->tocLineEnd();
3800                                 }
3801                         }
3802
3803                         $levelCount[$toclevel] = $level;
3804
3805                         # count number of headlines for each level
3806                         @$sublevelCount[$toclevel]++;
3807                         $dot = 0;
3808                         for( $i = 1; $i <= $toclevel; $i++ ) {
3809                                 if ( !empty( $sublevelCount[$i] ) ) {
3810                                         if ( $dot ) {
3811                                                 $numbering .= '.';
3812                                         }
3813                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3814                                         $dot = 1;
3815                                 }
3816                         }
3817
3818                         # The safe header is a version of the header text safe to use for links
3819                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3820                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3821
3822                         # Remove link placeholders by the link text.
3823                         #     <!--LINK number-->
3824                         # turns into
3825                         #     link text with suffix
3826                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3827
3828                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3829                         $tocline = preg_replace(
3830                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3831                                 array( '',                          '<$1>' ),
3832                                 $safeHeadline
3833                         );
3834                         $tocline = trim( $tocline );
3835
3836                         # For the anchor, strip out HTML-y stuff period
3837                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3838                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3839
3840                         # Save headline for section edit hint before it's escaped
3841                         $headlineHint = $safeHeadline;
3842
3843                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3844                                 # For reverse compatibility, provide an id that's
3845                                 # HTML4-compatible, like we used to.
3846                                 #
3847                                 # It may be worth noting, academically, that it's possible for
3848                                 # the legacy anchor to conflict with a non-legacy headline
3849                                 # anchor on the page.  In this case likely the "correct" thing
3850                                 # would be to either drop the legacy anchors or make sure
3851                                 # they're numbered first.  However, this would require people
3852                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3853                                 # manually, so let's not bother worrying about it.
3854                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3855                                         array( 'noninitial', 'legacy' ) );
3856                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3857
3858                                 if ( $legacyHeadline == $safeHeadline ) {
3859                                         # No reason to have both (in fact, we can't)
3860                                         $legacyHeadline = false;
3861                                 }
3862                         } else {
3863                                 $legacyHeadline = false;
3864                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3865                                         'noninitial' );
3866                         }
3867
3868                         # HTML names must be case-insensitively unique (bug 10721).
3869                         # This does not apply to Unicode characters per
3870                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3871                         # FIXME: We may be changing them depending on the current locale.
3872                         $arrayKey = strtolower( $safeHeadline );
3873                         if ( $legacyHeadline === false ) {
3874                                 $legacyArrayKey = false;
3875                         } else {
3876                                 $legacyArrayKey = strtolower( $legacyHeadline );
3877                         }
3878
3879                         # count how many in assoc. array so we can track dupes in anchors
3880                         if ( isset( $refers[$arrayKey] ) ) {
3881                                 $refers[$arrayKey]++;
3882                         } else {
3883                                 $refers[$arrayKey] = 1;
3884                         }
3885                         if ( isset( $refers[$legacyArrayKey] ) ) {
3886                                 $refers[$legacyArrayKey]++;
3887                         } else {
3888                                 $refers[$legacyArrayKey] = 1;
3889                         }
3890
3891                         # Don't number the heading if it is the only one (looks silly)
3892                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3893                                 # the two are different if the line contains a link
3894                                 $headline = $numbering . ' ' . $headline;
3895                         }
3896
3897                         # Create the anchor for linking from the TOC to the section
3898                         $anchor = $safeHeadline;
3899                         $legacyAnchor = $legacyHeadline;
3900                         if ( $refers[$arrayKey] > 1 ) {
3901                                 $anchor .= '_' . $refers[$arrayKey];
3902                         }
3903                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3904                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3905                         }
3906                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3907                                 $toc .= $sk->tocLine( $anchor, $tocline,
3908                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3909                         }
3910
3911                         # Add the section to the section tree
3912                         # Find the DOM node for this header
3913                         while ( $node && !$isTemplate ) {
3914                                 if ( $node->getName() === 'h' ) {
3915                                         $bits = $node->splitHeading();
3916                                         if ( $bits['i'] == $sectionIndex ) {
3917                                                 break;
3918                                         }
3919                                 }
3920                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3921                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3922                                 $node = $node->getNextSibling();
3923                         }
3924                         $tocraw[] = array(
3925                                 'toclevel' => $toclevel,
3926                                 'level' => $level,
3927                                 'line' => $tocline,
3928                                 'number' => $numbering,
3929                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3930                                 'fromtitle' => $titleText,
3931                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3932                                 'anchor' => $anchor,
3933                         );
3934
3935                         # give headline the correct <h#> tag
3936                         if ( $showEditLink && $sectionIndex !== false ) {
3937                                 // Output edit section links as markers with styles that can be customized by skins
3938                                 if ( $isTemplate ) {
3939                                         # Put a T flag in the section identifier, to indicate to extractSections()
3940                                         # that sections inside <includeonly> should be counted.
3941                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
3942                                 } else {
3943                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
3944                                 }
3945                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
3946                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
3947                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
3948                                 // so we don't have to worry about a user trying to input one of these markers directly.
3949                                 // We use a page and section attribute to stop the language converter from converting these important bits
3950                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
3951                                 // be able to convert that piece of data.
3952                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
3953                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
3954                                 if ( isset($editlinkArgs[2]) ) {
3955                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
3956                                 } else {
3957                                         $editlink .= '/>';
3958                                 }
3959                         } else {
3960                                 $editlink = '';
3961                         }
3962                         $head[$headlineCount] = $sk->makeHeadline( $level,
3963                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3964                                 $editlink, $legacyAnchor );
3965
3966                         $headlineCount++;
3967                 }
3968
3969                 $this->setOutputType( $oldType );
3970
3971                 # Never ever show TOC if no headers
3972                 if ( $numVisible < 1 ) {
3973                         $enoughToc = false;
3974                 }
3975
3976                 if ( $enoughToc ) {
3977                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3978                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3979                         }
3980                         $toc = $sk->tocList( $toc );
3981                         $this->mOutput->setTOCHTML( $toc );
3982                 }
3983
3984                 if ( $isMain ) {
3985                         $this->mOutput->setSections( $tocraw );
3986                 }
3987
3988                 # split up and insert constructed headlines
3989
3990                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3991                 $i = 0;
3992
3993                 foreach ( $blocks as $block ) {
3994                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3995                                 # This is the [edit] link that appears for the top block of text when
3996                                 # section editing is enabled
3997
3998                                 # Disabled because it broke block formatting
3999                                 # For example, a bullet point in the top line
4000                                 # $full .= $sk->editSectionLink(0);
4001                         }
4002                         $full .= $block;
4003                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4004                                 # Top anchor now in skin
4005                                 $full = $full.$toc;
4006                         }
4007
4008                         if ( !empty( $head[$i] ) ) {
4009                                 $full .= $head[$i];
4010                         }
4011                         $i++;
4012                 }
4013                 if ( $this->mForceTocPosition ) {
4014                         return str_replace( '<!--MWTOC-->', $toc, $full );
4015                 } else {
4016                         return $full;
4017                 }
4018         }
4019
4020         /**
4021          * Transform wiki markup when saving a page by doing \r\n -> \n
4022          * conversion, substitting signatures, {{subst:}} templates, etc.
4023          *
4024          * @param $text String: the text to transform
4025          * @param $title Title: the Title object for the current article
4026          * @param $user User: the User object describing the current user
4027          * @param $options ParserOptions: parsing options
4028          * @param $clearState Boolean: whether to clear the parser state first
4029          * @return String: the altered wiki markup
4030          */
4031         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4032                 $this->startExternalParse( $title, $options, self::OT_WIKI, $clearState );
4033                 $this->setUser( $user );
4034
4035                 $pairs = array(
4036                         "\r\n" => "\n",
4037                 );
4038                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4039                 if( $options->getPreSaveTransform() ) {
4040                         $text = $this->pstPass2( $text, $user );
4041                 }
4042                 $text = $this->mStripState->unstripBoth( $text );
4043
4044                 $this->setUser( null ); #Reset
4045
4046                 return $text;
4047         }
4048
4049         /**
4050          * Pre-save transform helper function
4051          * @private
4052          */
4053         function pstPass2( $text, $user ) {
4054                 global $wgContLang, $wgLocaltimezone;
4055
4056                 # Note: This is the timestamp saved as hardcoded wikitext to
4057                 # the database, we use $wgContLang here in order to give
4058                 # everyone the same signature and use the default one rather
4059                 # than the one selected in each user's preferences.
4060                 # (see also bug 12815)
4061                 $ts = $this->mOptions->getTimestamp();
4062                 if ( isset( $wgLocaltimezone ) ) {
4063                         $tz = $wgLocaltimezone;
4064                 } else {
4065                         $tz = date_default_timezone_get();
4066                 }
4067
4068                 $unixts = wfTimestamp( TS_UNIX, $ts );
4069                 $oldtz = date_default_timezone_get();
4070                 date_default_timezone_set( $tz );
4071                 $ts = date( 'YmdHis', $unixts );
4072                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4073
4074                 # Allow translation of timezones through wiki. date() can return
4075                 # whatever crap the system uses, localised or not, so we cannot
4076                 # ship premade translations.
4077                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4078                 $msg = wfMessage( $key )->inContentLanguage();
4079                 if ( $msg->exists() ) {
4080                         $tzMsg = $msg->text();
4081                 }
4082
4083                 date_default_timezone_set( $oldtz );
4084
4085                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4086
4087                 # Variable replacement
4088                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4089                 $text = $this->replaceVariables( $text );
4090
4091                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4092                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4093
4094                 # Signatures
4095                 $sigText = $this->getUserSig( $user );
4096                 $text = strtr( $text, array(
4097                         '~~~~~' => $d,
4098                         '~~~~' => "$sigText $d",
4099                         '~~~' => $sigText
4100                 ) );
4101
4102                 # Context links: [[|name]] and [[name (context)|]]
4103                 global $wgLegalTitleChars;
4104                 $tc = "[$wgLegalTitleChars]";
4105                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4106
4107                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4108                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4109                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4110                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4111
4112                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4113                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4114                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4115                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4116
4117                 $t = $this->mTitle->getText();
4118                 $m = array();
4119                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4120                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4121                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4122                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4123                 } else {
4124                         # if there's no context, don't bother duplicating the title
4125                         $text = preg_replace( $p2, '[[\\1]]', $text );
4126                 }
4127
4128                 # Trim trailing whitespace
4129                 $text = rtrim( $text );
4130
4131                 return $text;
4132         }
4133
4134         /**
4135          * Fetch the user's signature text, if any, and normalize to
4136          * validated, ready-to-insert wikitext.
4137          * If you have pre-fetched the nickname or the fancySig option, you can
4138          * specify them here to save a database query.
4139          * Do not reuse this parser instance after calling getUserSig(),
4140          * as it may have changed if it's the $wgParser.
4141          *
4142          * @param $user User
4143          * @param $nickname String: nickname to use or false to use user's default nickname
4144          * @param $fancySig Boolean: whether the nicknname is the complete signature
4145          *                  or null to use default value
4146          * @return string
4147          */
4148         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4149                 global $wgMaxSigChars;
4150
4151                 $username = $user->getName();
4152
4153                 # If not given, retrieve from the user object.
4154                 if ( $nickname === false )
4155                         $nickname = $user->getOption( 'nickname' );
4156
4157                 if ( is_null( $fancySig ) ) {
4158                         $fancySig = $user->getBoolOption( 'fancysig' );
4159                 }
4160
4161                 $nickname = $nickname == null ? $username : $nickname;
4162
4163                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4164                         $nickname = $username;
4165                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4166                 } elseif ( $fancySig !== false ) {
4167                         # Sig. might contain markup; validate this
4168                         if ( $this->validateSig( $nickname ) !== false ) {
4169                                 # Validated; clean up (if needed) and return it
4170                                 return $this->cleanSig( $nickname, true );
4171                         } else {
4172                                 # Failed to validate; fall back to the default
4173                                 $nickname = $username;
4174                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4175                         }
4176                 }
4177
4178                 # Make sure nickname doesnt get a sig in a sig
4179                 $nickname = $this->cleanSigInSig( $nickname );
4180
4181                 # If we're still here, make it a link to the user page
4182                 $userText = wfEscapeWikiText( $username );
4183                 $nickText = wfEscapeWikiText( $nickname );
4184                 if ( $user->isAnon() )  {
4185                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4186                 } else {
4187                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4188                 }
4189         }
4190
4191         /**
4192          * Check that the user's signature contains no bad XML
4193          *
4194          * @param $text String
4195          * @return mixed An expanded string, or false if invalid.
4196          */
4197         function validateSig( $text ) {
4198                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4199         }
4200
4201         /**
4202          * Clean up signature text
4203          *
4204          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4205          * 2) Substitute all transclusions
4206          *
4207          * @param $text String
4208          * @param $parsing Whether we're cleaning (preferences save) or parsing
4209          * @return String: signature text
4210          */
4211         function cleanSig( $text, $parsing = false ) {
4212                 if ( !$parsing ) {
4213                         global $wgTitle;
4214                         $this->mOptions = new ParserOptions;
4215                         $this->clearState();
4216                         $this->setTitle( $wgTitle );
4217                         $this->setOutputType = self::OT_PREPROCESS;
4218                 }
4219
4220                 # Option to disable this feature
4221                 if ( !$this->mOptions->getCleanSignatures() ) {
4222                         return $text;
4223                 }
4224
4225                 # FIXME: regex doesn't respect extension tags or nowiki
4226                 #  => Move this logic to braceSubstitution()
4227                 $substWord = MagicWord::get( 'subst' );
4228                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4229                 $substText = '{{' . $substWord->getSynonym( 0 );
4230
4231                 $text = preg_replace( $substRegex, $substText, $text );
4232                 $text = $this->cleanSigInSig( $text );
4233                 $dom = $this->preprocessToDom( $text );
4234                 $frame = $this->getPreprocessor()->newFrame();
4235                 $text = $frame->expand( $dom );
4236
4237                 if ( !$parsing ) {
4238                         $text = $this->mStripState->unstripBoth( $text );
4239                 }
4240
4241                 return $text;
4242         }
4243
4244         /**
4245          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4246          *
4247          * @param $text String
4248          * @return String: signature text with /~{3,5}/ removed
4249          */
4250         function cleanSigInSig( $text ) {
4251                 $text = preg_replace( '/~{3,5}/', '', $text );
4252                 return $text;
4253         }
4254
4255         /**
4256          * Set up some variables which are usually set up in parse()
4257          * so that an external function can call some class members with confidence
4258          */
4259         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4260                 $this->setTitle( $title );
4261                 $this->mOptions = $options;
4262                 $this->setOutputType( $outputType );
4263                 if ( $clearState ) {
4264                         $this->clearState();
4265                 }
4266         }
4267
4268         /**
4269          * Wrapper for preprocess()
4270          *
4271          * @param $text String: the text to preprocess
4272          * @param $options ParserOptions: options
4273          * @param $title Title object or null to use $wgTitle
4274          * @return String
4275          */
4276         public function transformMsg( $text, $options, $title = null ) {
4277                 static $executing = false;
4278
4279                 # Guard against infinite recursion
4280                 if ( $executing ) {
4281                         return $text;
4282                 }
4283                 $executing = true;
4284
4285                 wfProfileIn( __METHOD__ );
4286                 if ( !$title ) {
4287                         global $wgTitle;
4288                         $title = $wgTitle;
4289                 }
4290                 if ( !$title ) {
4291                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4292                         # Create a ghost title in such case
4293                         $title = Title::newFromText( 'Dwimmerlaik' );
4294                 }
4295                 $text = $this->preprocess( $text, $title, $options );
4296
4297                 $executing = false;
4298                 wfProfileOut( __METHOD__ );
4299                 return $text;
4300         }
4301
4302         /**
4303          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4304          * The callback should have the following form:
4305          *    function myParserHook( $text, $params, $parser ) { ... }
4306          *
4307          * Transform and return $text. Use $parser for any required context, e.g. use
4308          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4309          *
4310          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4311          * @param $callback Mixed: the callback function (and object) to use for the tag
4312          * @return The old value of the mTagHooks array associated with the hook
4313          */
4314         public function setHook( $tag, $callback ) {
4315                 $tag = strtolower( $tag );
4316                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4317                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4318                 $this->mTagHooks[$tag] = $callback;
4319                 if ( !in_array( $tag, $this->mStripList ) ) {
4320                         $this->mStripList[] = $tag;
4321                 }
4322
4323                 return $oldVal;
4324         }
4325
4326         function setTransparentTagHook( $tag, $callback ) {
4327                 $tag = strtolower( $tag );
4328                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4329                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4330                 $this->mTransparentTagHooks[$tag] = $callback;
4331
4332                 return $oldVal;
4333         }
4334
4335         /**
4336          * Remove all tag hooks
4337          */
4338         function clearTagHooks() {
4339                 $this->mTagHooks = array();
4340                 $this->mStripList = $this->mDefaultStripList;
4341         }
4342
4343         /**
4344          * Create a function, e.g. {{sum:1|2|3}}
4345          * The callback function should have the form:
4346          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4347          *
4348          * Or with SFH_OBJECT_ARGS:
4349          *    function myParserFunction( $parser, $frame, $args ) { ... }
4350          *
4351          * The callback may either return the text result of the function, or an array with the text
4352          * in element 0, and a number of flags in the other elements. The names of the flags are
4353          * specified in the keys. Valid flags are:
4354          *   found                     The text returned is valid, stop processing the template. This
4355          *                             is on by default.
4356          *   nowiki                    Wiki markup in the return value should be escaped
4357          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4358          *
4359          * @param $id String: The magic word ID
4360          * @param $callback Mixed: the callback function (and object) to use
4361          * @param $flags Integer: a combination of the following flags:
4362          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4363          *
4364          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4365          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4366          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4367          *     the arguments, and to control the way they are expanded.
4368          *
4369          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4370          *     arguments, for instance:
4371          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4372          *
4373          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4374          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4375          *     working if/when this is changed.
4376          *
4377          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4378          *     expansion.
4379          *
4380          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4381          *     about the methods available in PPFrame and PPNode.
4382          *
4383          * @return The old callback function for this name, if any
4384          */
4385         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4386                 global $wgContLang;
4387
4388                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4389                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4390
4391                 # Add to function cache
4392                 $mw = MagicWord::get( $id );
4393                 if ( !$mw )
4394                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4395
4396                 $synonyms = $mw->getSynonyms();
4397                 $sensitive = intval( $mw->isCaseSensitive() );
4398
4399                 foreach ( $synonyms as $syn ) {
4400                         # Case
4401                         if ( !$sensitive ) {
4402                                 $syn = $wgContLang->lc( $syn );
4403                         }
4404                         # Add leading hash
4405                         if ( !( $flags & SFH_NO_HASH ) ) {
4406                                 $syn = '#' . $syn;
4407                         }
4408                         # Remove trailing colon
4409                         if ( substr( $syn, -1, 1 ) === ':' ) {
4410                                 $syn = substr( $syn, 0, -1 );
4411                         }
4412                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4413                 }
4414                 return $oldVal;
4415         }
4416
4417         /**
4418          * Get all registered function hook identifiers
4419          *
4420          * @return Array
4421          */
4422         function getFunctionHooks() {
4423                 return array_keys( $this->mFunctionHooks );
4424         }
4425
4426         /**
4427          * Create a tag function, e.g. <test>some stuff</test>.
4428          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4429          * Unlike parser functions, their content is not preprocessed.
4430          */
4431         function setFunctionTagHook( $tag, $callback, $flags ) {
4432                 $tag = strtolower( $tag );
4433                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4434                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4435                         $this->mFunctionTagHooks[$tag] : null;
4436                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4437
4438                 if ( !in_array( $tag, $this->mStripList ) ) {
4439                         $this->mStripList[] = $tag;
4440                 }
4441
4442                 return $old;
4443         }
4444
4445         /**
4446          * FIXME: update documentation. makeLinkObj() is deprecated.
4447          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4448          * Placeholders created in Skin::makeLinkObj()
4449          * Returns an array of link CSS classes, indexed by PDBK.
4450          */
4451         function replaceLinkHolders( &$text, $options = 0 ) {
4452                 return $this->mLinkHolders->replace( $text );
4453         }
4454
4455         /**
4456          * Replace <!--LINK--> link placeholders with plain text of links
4457          * (not HTML-formatted).
4458          *
4459          * @param $text String
4460          * @return String
4461          */
4462         function replaceLinkHoldersText( $text ) {
4463                 return $this->mLinkHolders->replaceText( $text );
4464         }
4465
4466         /**
4467          * Renders an image gallery from a text with one line per image.
4468          * text labels may be given by using |-style alternative text. E.g.
4469          *   Image:one.jpg|The number "1"
4470          *   Image:tree.jpg|A tree
4471          * given as text will return the HTML of a gallery with two images,
4472          * labeled 'The number "1"' and
4473          * 'A tree'.
4474          */
4475         function renderImageGallery( $text, $params ) {
4476                 $ig = new ImageGallery();
4477                 $ig->setContextTitle( $this->mTitle );
4478                 $ig->setShowBytes( false );
4479                 $ig->setShowFilename( false );
4480                 $ig->setParser( $this );
4481                 $ig->setHideBadImages();
4482                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4483                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4484                 $ig->mRevisionId = $this->mRevisionId;
4485
4486                 if ( isset( $params['showfilename'] ) ) {
4487                         $ig->setShowFilename( true );
4488                 } else {
4489                         $ig->setShowFilename( false );
4490                 }
4491                 if ( isset( $params['caption'] ) ) {
4492                         $caption = $params['caption'];
4493                         $caption = htmlspecialchars( $caption );
4494                         $caption = $this->replaceInternalLinks( $caption );
4495                         $ig->setCaptionHtml( $caption );
4496                 }
4497                 if ( isset( $params['perrow'] ) ) {
4498                         $ig->setPerRow( $params['perrow'] );
4499                 }
4500                 if ( isset( $params['widths'] ) ) {
4501                         $ig->setWidths( $params['widths'] );
4502                 }
4503                 if ( isset( $params['heights'] ) ) {
4504                         $ig->setHeights( $params['heights'] );
4505                 }
4506
4507                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4508
4509                 $lines = StringUtils::explode( "\n", $text );
4510                 foreach ( $lines as $line ) {
4511                         # match lines like these:
4512                         # Image:someimage.jpg|This is some image
4513                         $matches = array();
4514                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4515                         # Skip empty lines
4516                         if ( count( $matches ) == 0 ) {
4517                                 continue;
4518                         }
4519
4520                         if ( strpos( $matches[0], '%' ) !== false ) {
4521                                 $matches[1] = rawurldecode( $matches[1] );
4522                         }
4523                         $tp = Title::newFromText( $matches[1], NS_FILE );
4524                         $nt =& $tp;
4525                         if ( is_null( $nt ) ) {
4526                                 # Bogus title. Ignore these so we don't bomb out later.
4527                                 continue;
4528                         }
4529                         if ( isset( $matches[3] ) ) {
4530                                 $label = $matches[3];
4531                         } else {
4532                                 $label = '';
4533                         }
4534
4535                         $html = $this->recursiveTagParse( trim( $label ) );
4536
4537                         $ig->add( $nt, $html );
4538
4539                         # Only add real images (bug #5586)
4540                         if ( $nt->getNamespace() == NS_FILE ) {
4541                                 $this->mOutput->addImage( $nt->getDBkey() );
4542                         }
4543                 }
4544                 return $ig->toHTML();
4545         }
4546
4547         function getImageParams( $handler ) {
4548                 if ( $handler ) {
4549                         $handlerClass = get_class( $handler );
4550                 } else {
4551                         $handlerClass = '';
4552                 }
4553                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4554                         # Initialise static lists
4555                         static $internalParamNames = array(
4556                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4557                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4558                                         'bottom', 'text-bottom' ),
4559                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4560                                         'upright', 'border', 'link', 'alt' ),
4561                         );
4562                         static $internalParamMap;
4563                         if ( !$internalParamMap ) {
4564                                 $internalParamMap = array();
4565                                 foreach ( $internalParamNames as $type => $names ) {
4566                                         foreach ( $names as $name ) {
4567                                                 $magicName = str_replace( '-', '_', "img_$name" );
4568                                                 $internalParamMap[$magicName] = array( $type, $name );
4569                                         }
4570                                 }
4571                         }
4572
4573                         # Add handler params
4574                         $paramMap = $internalParamMap;
4575                         if ( $handler ) {
4576                                 $handlerParamMap = $handler->getParamMap();
4577                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4578                                         $paramMap[$magic] = array( 'handler', $paramName );
4579                                 }
4580                         }
4581                         $this->mImageParams[$handlerClass] = $paramMap;
4582                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4583                 }
4584                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4585         }
4586
4587         /**
4588          * Parse image options text and use it to make an image
4589          *
4590          * @param $title Title
4591          * @param $options String
4592          * @param $holders LinkHolderArray
4593          */
4594         function makeImage( $title, $options, $holders = false ) {
4595                 # Check if the options text is of the form "options|alt text"
4596                 # Options are:
4597                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4598                 #  * left       no resizing, just left align. label is used for alt= only
4599                 #  * right      same, but right aligned
4600                 #  * none       same, but not aligned
4601                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4602                 #  * center     center the image
4603                 #  * frame      Keep original image size, no magnify-button.
4604                 #  * framed     Same as "frame"
4605                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4606                 #  * upright    reduce width for upright images, rounded to full __0 px
4607                 #  * border     draw a 1px border around the image
4608                 #  * alt        Text for HTML alt attribute (defaults to empty)
4609                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4610                 # vertical-align values (no % or length right now):
4611                 #  * baseline
4612                 #  * sub
4613                 #  * super
4614                 #  * top
4615                 #  * text-top
4616                 #  * middle
4617                 #  * bottom
4618                 #  * text-bottom
4619
4620                 $parts = StringUtils::explode( "|", $options );
4621                 $sk = $this->mOptions->getSkin( $this->mTitle );
4622
4623                 # Give extensions a chance to select the file revision for us
4624                 $skip = $time = $descQuery = false;
4625                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4626
4627                 if ( $skip ) {
4628                         return $sk->link( $title );
4629                 }
4630
4631                 # Get the file
4632                 $file = wfFindFile( $title, array( 'time' => $time ) );
4633                 # Get parameter map
4634                 $handler = $file ? $file->getHandler() : false;
4635
4636                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4637
4638                 # Process the input parameters
4639                 $caption = '';
4640                 $params = array( 'frame' => array(), 'handler' => array(),
4641                         'horizAlign' => array(), 'vertAlign' => array() );
4642                 foreach ( $parts as $part ) {
4643                         $part = trim( $part );
4644                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4645                         $validated = false;
4646                         if ( isset( $paramMap[$magicName] ) ) {
4647                                 list( $type, $paramName ) = $paramMap[$magicName];
4648
4649                                 # Special case; width and height come in one variable together
4650                                 if ( $type === 'handler' && $paramName === 'width' ) {
4651                                         $m = array();
4652                                         # (bug 13500) In both cases (width/height and width only),
4653                                         # permit trailing "px" for backward compatibility.
4654                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4655                                                 $width = intval( $m[1] );
4656                                                 $height = intval( $m[2] );
4657                                                 if ( $handler->validateParam( 'width', $width ) ) {
4658                                                         $params[$type]['width'] = $width;
4659                                                         $validated = true;
4660                                                 }
4661                                                 if ( $handler->validateParam( 'height', $height ) ) {
4662                                                         $params[$type]['height'] = $height;
4663                                                         $validated = true;
4664                                                 }
4665                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4666                                                 $width = intval( $value );
4667                                                 if ( $handler->validateParam( 'width', $width ) ) {
4668                                                         $params[$type]['width'] = $width;
4669                                                         $validated = true;
4670                                                 }
4671                                         } # else no validation -- bug 13436
4672                                 } else {
4673                                         if ( $type === 'handler' ) {
4674                                                 # Validate handler parameter
4675                                                 $validated = $handler->validateParam( $paramName, $value );
4676                                         } else {
4677                                                 # Validate internal parameters
4678                                                 switch( $paramName ) {
4679                                                 case 'manualthumb':
4680                                                 case 'alt':
4681                                                         # @todo Fixme: possibly check validity here for
4682                                                         # manualthumb? downstream behavior seems odd with
4683                                                         # missing manual thumbs.
4684                                                         $validated = true;
4685                                                         $value = $this->stripAltText( $value, $holders );
4686                                                         break;
4687                                                 case 'link':
4688                                                         $chars = self::EXT_LINK_URL_CLASS;
4689                                                         $prots = $this->mUrlProtocols;
4690                                                         if ( $value === '' ) {
4691                                                                 $paramName = 'no-link';
4692                                                                 $value = true;
4693                                                                 $validated = true;
4694                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4695                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4696                                                                         $paramName = 'link-url';
4697                                                                         $this->mOutput->addExternalLink( $value );
4698                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4699                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4700                                                                         }
4701                                                                         $validated = true;
4702                                                                 }
4703                                                         } else {
4704                                                                 $linkTitle = Title::newFromText( $value );
4705                                                                 if ( $linkTitle ) {
4706                                                                         $paramName = 'link-title';
4707                                                                         $value = $linkTitle;
4708                                                                         $this->mOutput->addLink( $linkTitle );
4709                                                                         $validated = true;
4710                                                                 }
4711                                                         }
4712                                                         break;
4713                                                 default:
4714                                                         # Most other things appear to be empty or numeric...
4715                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4716                                                 }
4717                                         }
4718
4719                                         if ( $validated ) {
4720                                                 $params[$type][$paramName] = $value;
4721                                         }
4722                                 }
4723                         }
4724                         if ( !$validated ) {
4725                                 $caption = $part;
4726                         }
4727                 }
4728
4729                 # Process alignment parameters
4730                 if ( $params['horizAlign'] ) {
4731                         $params['frame']['align'] = key( $params['horizAlign'] );
4732                 }
4733                 if ( $params['vertAlign'] ) {
4734                         $params['frame']['valign'] = key( $params['vertAlign'] );
4735                 }
4736
4737                 $params['frame']['caption'] = $caption;
4738
4739                 # Will the image be presented in a frame, with the caption below?
4740                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4741                                                  isset( $params['frame']['framed'] ) ||
4742                                                  isset( $params['frame']['thumbnail'] ) ||
4743                                                  isset( $params['frame']['manualthumb'] );
4744
4745                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4746                 # came to also set the caption, ordinary text after the image -- which
4747                 # makes no sense, because that just repeats the text multiple times in
4748                 # screen readers.  It *also* came to set the title attribute.
4749                 #
4750                 # Now that we have an alt attribute, we should not set the alt text to
4751                 # equal the caption: that's worse than useless, it just repeats the
4752                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4753                 # use the unnamed parameter for alt text as well, just for the time be-
4754                 # ing, if the unnamed param is set and the alt param is not.
4755                 #
4756                 # For the future, we need to figure out if we want to tweak this more,
4757                 # e.g., introducing a title= parameter for the title; ignoring the un-
4758                 # named parameter entirely for images without a caption; adding an ex-
4759                 # plicit caption= parameter and preserving the old magic unnamed para-
4760                 # meter for BC; ...
4761                 if ( $imageIsFramed ) { # Framed image
4762                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4763                                 # No caption or alt text, add the filename as the alt text so
4764                                 # that screen readers at least get some description of the image
4765                                 $params['frame']['alt'] = $title->getText();
4766                         }
4767                         # Do not set $params['frame']['title'] because tooltips don't make sense
4768                         # for framed images
4769                 } else { # Inline image
4770                         if ( !isset( $params['frame']['alt'] ) ) {
4771                                 # No alt text, use the "caption" for the alt text
4772                                 if ( $caption !== '') {
4773                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4774                                 } else {
4775                                         # No caption, fall back to using the filename for the
4776                                         # alt text
4777                                         $params['frame']['alt'] = $title->getText();
4778                                 }
4779                         }
4780                         # Use the "caption" for the tooltip text
4781                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4782                 }
4783
4784                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4785
4786                 # Linker does the rest
4787                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4788
4789                 # Give the handler a chance to modify the parser object
4790                 if ( $handler ) {
4791                         $handler->parserTransformHook( $this, $file );
4792                 }
4793
4794                 return $ret;
4795         }
4796
4797         protected function stripAltText( $caption, $holders ) {
4798                 # Strip bad stuff out of the title (tooltip).  We can't just use
4799                 # replaceLinkHoldersText() here, because if this function is called
4800                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4801                 if ( $holders ) {
4802                         $tooltip = $holders->replaceText( $caption );
4803                 } else {
4804                         $tooltip = $this->replaceLinkHoldersText( $caption );
4805                 }
4806
4807                 # make sure there are no placeholders in thumbnail attributes
4808                 # that are later expanded to html- so expand them now and
4809                 # remove the tags
4810                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4811                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4812
4813                 return $tooltip;
4814         }
4815
4816         /**
4817          * Set a flag in the output object indicating that the content is dynamic and
4818          * shouldn't be cached.
4819          */
4820         function disableCache() {
4821                 wfDebug( "Parser output marked as uncacheable.\n" );
4822                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4823                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4824         }
4825
4826         /**
4827          * Callback from the Sanitizer for expanding items found in HTML attribute
4828          * values, so they can be safely tested and escaped.
4829          *
4830          * @param $text String
4831          * @param $frame PPFrame
4832          * @return String
4833          * @private
4834          */
4835         function attributeStripCallback( &$text, $frame = false ) {
4836                 $text = $this->replaceVariables( $text, $frame );
4837                 $text = $this->mStripState->unstripBoth( $text );
4838                 return $text;
4839         }
4840
4841         /**
4842          * Accessor
4843          */
4844         function getTags() {
4845                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4846         }
4847
4848         /**
4849          * Break wikitext input into sections, and either pull or replace
4850          * some particular section's text.
4851          *
4852          * External callers should use the getSection and replaceSection methods.
4853          *
4854          * @param $text String: Page wikitext
4855          * @param $section String: a section identifier string of the form:
4856          *   <flag1> - <flag2> - ... - <section number>
4857          *
4858          * Currently the only recognised flag is "T", which means the target section number
4859          * was derived during a template inclusion parse, in other words this is a template
4860          * section edit link. If no flags are given, it was an ordinary section edit link.
4861          * This flag is required to avoid a section numbering mismatch when a section is
4862          * enclosed by <includeonly> (bug 6563).
4863          *
4864          * The section number 0 pulls the text before the first heading; other numbers will
4865          * pull the given section along with its lower-level subsections. If the section is
4866          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4867          *
4868          * @param $mode String: one of "get" or "replace"
4869          * @param $newText String: replacement text for section data.
4870          * @return String: for "get", the extracted section text.
4871          *                 for "replace", the whole page with the section replaced.
4872          */
4873         private function extractSections( $text, $section, $mode, $newText='' ) {
4874                 global $wgTitle; # not generally used but removes an ugly failure mode
4875                 $this->startExternalParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
4876                 $outText = '';
4877                 $frame = $this->getPreprocessor()->newFrame();
4878
4879                 # Process section extraction flags
4880                 $flags = 0;
4881                 $sectionParts = explode( '-', $section );
4882                 $sectionIndex = array_pop( $sectionParts );
4883                 foreach ( $sectionParts as $part ) {
4884                         if ( $part === 'T' ) {
4885                                 $flags |= self::PTD_FOR_INCLUSION;
4886                         }
4887                 }
4888                 # Preprocess the text
4889                 $root = $this->preprocessToDom( $text, $flags );
4890
4891                 # <h> nodes indicate section breaks
4892                 # They can only occur at the top level, so we can find them by iterating the root's children
4893                 $node = $root->getFirstChild();
4894
4895                 # Find the target section
4896                 if ( $sectionIndex == 0 ) {
4897                         # Section zero doesn't nest, level=big
4898                         $targetLevel = 1000;
4899                 } else {
4900                         while ( $node ) {
4901                                 if ( $node->getName() === 'h' ) {
4902                                         $bits = $node->splitHeading();
4903                                         if ( $bits['i'] == $sectionIndex ) {
4904                                                 $targetLevel = $bits['level'];
4905                                                 break;
4906                                         }
4907                                 }
4908                                 if ( $mode === 'replace' ) {
4909                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4910                                 }
4911                                 $node = $node->getNextSibling();
4912                         }
4913                 }
4914
4915                 if ( !$node ) {
4916                         # Not found
4917                         if ( $mode === 'get' ) {
4918                                 return $newText;
4919                         } else {
4920                                 return $text;
4921                         }
4922                 }
4923
4924                 # Find the end of the section, including nested sections
4925                 do {
4926                         if ( $node->getName() === 'h' ) {
4927                                 $bits = $node->splitHeading();
4928                                 $curLevel = $bits['level'];
4929                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4930                                         break;
4931                                 }
4932                         }
4933                         if ( $mode === 'get' ) {
4934                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4935                         }
4936                         $node = $node->getNextSibling();
4937                 } while ( $node );
4938
4939                 # Write out the remainder (in replace mode only)
4940                 if ( $mode === 'replace' ) {
4941                         # Output the replacement text
4942                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4943                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4944                         # Only add trailing whitespace if there is newText
4945                         if ( $newText != "" ) {
4946                                 $outText .= $newText . "\n\n";
4947                         }
4948
4949                         while ( $node ) {
4950                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4951                                 $node = $node->getNextSibling();
4952                         }
4953                 }
4954
4955                 if ( is_string( $outText ) ) {
4956                         # Re-insert stripped tags
4957                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4958                 }
4959
4960                 return $outText;
4961         }
4962
4963         /**
4964          * This function returns the text of a section, specified by a number ($section).
4965          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4966          * the first section before any such heading (section 0).
4967          *
4968          * If a section contains subsections, these are also returned.
4969          *
4970          * @param $text String: text to look in
4971          * @param $section String: section identifier
4972          * @param $deftext String: default to return if section is not found
4973          * @return string text of the requested section
4974          */
4975         public function getSection( $text, $section, $deftext='' ) {
4976                 return $this->extractSections( $text, $section, "get", $deftext );
4977         }
4978
4979         /**
4980          * This function returns $oldtext after the content of the section
4981          * specified by $section has been replaced with $text.
4982          *
4983          * @param $text String: former text of the article
4984          * @param $section Numeric: section identifier
4985          * @param $text String: replacing text
4986          * #return String: modified text
4987          */
4988         public function replaceSection( $oldtext, $section, $text ) {
4989                 return $this->extractSections( $oldtext, $section, "replace", $text );
4990         }
4991
4992         /**
4993          * Get the ID of the revision we are parsing
4994          *
4995          * @return Mixed: integer or null
4996          */
4997         function getRevisionId() {
4998                 return $this->mRevisionId;
4999         }
5000
5001         /**
5002          * Get the revision object for $this->mRevisionId
5003          *
5004          * @return either a Revision object or null
5005          */
5006         protected function getRevisionObject() {
5007                 if ( !is_null( $this->mRevisionObject ) ) {
5008                         return $this->mRevisionObject;
5009                 }
5010                 if ( is_null( $this->mRevisionId ) ) {
5011                         return null;
5012                 }
5013
5014                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5015                 return $this->mRevisionObject;
5016         }
5017
5018         /**
5019          * Get the timestamp associated with the current revision, adjusted for
5020          * the default server-local timestamp
5021          */
5022         function getRevisionTimestamp() {
5023                 if ( is_null( $this->mRevisionTimestamp ) ) {
5024                         wfProfileIn( __METHOD__ );
5025
5026                         $revObject = $this->getRevisionObject();
5027                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5028
5029                         if( $timestamp !== false ) {
5030                                 global $wgContLang;
5031
5032                                 # The cryptic '' timezone parameter tells to use the site-default
5033                                 # timezone offset instead of the user settings.
5034                                 #
5035                                 # Since this value will be saved into the parser cache, served
5036                                 # to other users, and potentially even used inside links and such,
5037                                 # it needs to be consistent for all visitors.
5038                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5039                         }
5040
5041                         wfProfileOut( __METHOD__ );
5042                 }
5043                 return $this->mRevisionTimestamp;
5044         }
5045
5046         /**
5047          * Get the name of the user that edited the last revision
5048          *
5049          * @return String: user name
5050          */
5051         function getRevisionUser() {
5052                 if( is_null( $this->mRevisionUser ) ) {
5053                         $revObject = $this->getRevisionObject();
5054
5055                         # if this template is subst: the revision id will be blank,
5056                         # so just use the current user's name
5057                         if( $revObject ) {
5058                                 $this->mRevisionUser = $revObject->getUserText();
5059                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5060                                 $this->mRevisionUser = $this->getUser()->getName();
5061                         }
5062                 }
5063                 return $this->mRevisionUser;
5064         }
5065
5066         /**
5067          * Mutator for $mDefaultSort
5068          *
5069          * @param $sort New value
5070          */
5071         public function setDefaultSort( $sort ) {
5072                 $this->mDefaultSort = $sort;
5073                 $this->mOutput->setProperty( 'defaultsort', $sort );
5074         }
5075
5076         /**
5077          * Accessor for $mDefaultSort
5078          * Will use the empty string if none is set.
5079          *
5080          * This value is treated as a prefix, so the
5081          * empty string is equivalent to sorting by
5082          * page name.
5083          *
5084          * @return string
5085          */
5086         public function getDefaultSort() {
5087                 if ( $this->mDefaultSort !== false ) {
5088                         return $this->mDefaultSort;
5089                 } else {
5090                         return '';
5091                 }
5092         }
5093
5094         /**
5095          * Accessor for $mDefaultSort
5096          * Unlike getDefaultSort(), will return false if none is set
5097          *
5098          * @return string or false
5099          */
5100         public function getCustomDefaultSort() {
5101                 return $this->mDefaultSort;
5102         }
5103
5104         /**
5105          * Try to guess the section anchor name based on a wikitext fragment
5106          * presumably extracted from a heading, for example "Header" from
5107          * "== Header ==".
5108          */
5109         public function guessSectionNameFromWikiText( $text ) {
5110                 # Strip out wikitext links(they break the anchor)
5111                 $text = $this->stripSectionName( $text );
5112                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5113                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5114         }
5115
5116         /**
5117          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5118          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5119          * as something other than UTF-8 (apparently?), resulting in breakage.
5120          *
5121          * @param $text String: The section name
5122          * @return string An anchor
5123          */
5124         public function guessLegacySectionNameFromWikiText( $text ) {
5125                 # Strip out wikitext links(they break the anchor)
5126                 $text = $this->stripSectionName( $text );
5127                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5128                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5129         }
5130
5131         /**
5132          * Strips a text string of wikitext for use in a section anchor
5133          *
5134          * Accepts a text string and then removes all wikitext from the
5135          * string and leaves only the resultant text (i.e. the result of
5136          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5137          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5138          * to create valid section anchors by mimicing the output of the
5139          * parser when headings are parsed.
5140          *
5141          * @param $text String: text string to be stripped of wikitext
5142          * for use in a Section anchor
5143          * @return Filtered text string
5144          */
5145         public function stripSectionName( $text ) {
5146                 # Strip internal link markup
5147                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5148                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5149
5150                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5151                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5152                 # on how many empty links there are on the page - need to figure that out.
5153                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5154
5155                 # Parse wikitext quotes (italics & bold)
5156                 $text = $this->doQuotes( $text );
5157
5158                 # Strip HTML tags
5159                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5160                 return $text;
5161         }
5162
5163         /**
5164          * strip/replaceVariables/unstrip for preprocessor regression testing
5165          */
5166         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5167                 if ( !$title instanceof Title ) {
5168                         $title = Title::newFromText( $title );
5169                 }
5170                 $this->startExternalParse( $title, $options, $outputType, true );
5171
5172                 $text = $this->replaceVariables( $text );
5173                 $text = $this->mStripState->unstripBoth( $text );
5174                 $text = Sanitizer::removeHTMLtags( $text );
5175                 return $text;
5176         }
5177
5178         function testPst( $text, $title, $options ) {
5179                 global $wgUser;
5180                 if ( !$title instanceof Title ) {
5181                         $title = Title::newFromText( $title );
5182                 }
5183                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5184         }
5185
5186         function testPreprocess( $text, $title, $options ) {
5187                 if ( !$title instanceof Title ) {
5188                         $title = Title::newFromText( $title );
5189                 }
5190                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5191         }
5192
5193         function markerSkipCallback( $s, $callback ) {
5194                 $i = 0;
5195                 $out = '';
5196                 while ( $i < strlen( $s ) ) {
5197                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5198                         if ( $markerStart === false ) {
5199                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5200                                 break;
5201                         } else {
5202                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5203                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5204                                 if ( $markerEnd === false ) {
5205                                         $out .= substr( $s, $markerStart );
5206                                         break;
5207                                 } else {
5208                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5209                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5210                                         $i = $markerEnd;
5211                                 }
5212                         }
5213                 }
5214                 return $out;
5215         }
5216
5217         function serialiseHalfParsedText( $text ) {
5218                 $data = array();
5219                 $data['text'] = $text;
5220
5221                 # First, find all strip markers, and store their
5222                 #  data in an array.
5223                 $stripState = new StripState;
5224                 $pos = 0;
5225                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5226                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5227                 {
5228                         $end_pos += strlen( self::MARKER_SUFFIX );
5229                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5230
5231                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5232                                 $replaceArray = $stripState->general;
5233                                 $stripText = $this->mStripState->general->data[$marker];
5234                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5235                                 $replaceArray = $stripState->nowiki;
5236                                 $stripText = $this->mStripState->nowiki->data[$marker];
5237                         } else {
5238                                 throw new MWException( "Hanging strip marker: '$marker'." );
5239                         }
5240
5241                         $replaceArray->setPair( $marker, $stripText );
5242                         $pos = $end_pos;
5243                 }
5244                 $data['stripstate'] = $stripState;
5245
5246                 # Now, find all of our links, and store THEIR
5247                 #  data in an array! :)
5248                 $links = array( 'internal' => array(), 'interwiki' => array() );
5249                 $pos = 0;
5250
5251                 # Internal links
5252                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5253                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5254
5255                         $ns = trim( $ns );
5256                         if ( empty( $links['internal'][$ns] ) ) {
5257                                 $links['internal'][$ns] = array();
5258                         }
5259
5260                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5261                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5262                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5263                 }
5264
5265                 $pos = 0;
5266
5267                 # Interwiki links
5268                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5269                         $data = substr( $text, $start_pos );
5270                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5271                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5272                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5273                 }
5274
5275                 $data['linkholder'] = $links;
5276
5277                 return $data;
5278         }
5279
5280         /**
5281          * TODO: document
5282          * @param $data Array
5283          * @param $intPrefix String unique identifying prefix
5284          * @return String
5285          */
5286         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5287                 if ( !$intPrefix ) {
5288                         $intPrefix = self::getRandomString();
5289                 }
5290
5291                 # First, extract the strip state.
5292                 $stripState = $data['stripstate'];
5293                 $this->mStripState->general->merge( $stripState->general );
5294                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5295
5296                 # Now, extract the text, and renumber links
5297                 $text = $data['text'];
5298                 $links = $data['linkholder'];
5299
5300                 # Internal...
5301                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5302                         foreach ( $nsLinks as $key => $entry ) {
5303                                 $newKey = $intPrefix . '-' . $key;
5304                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5305
5306                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5307                         }
5308                 }
5309
5310                 # Interwiki...
5311                 foreach ( $links['interwiki'] as $key => $entry ) {
5312                         $newKey = "$intPrefix-$key";
5313                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5314
5315                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5316                 }
5317
5318                 # Should be good to go.
5319                 return $text;
5320         }
5321 }
5322
5323 /**
5324  * @todo document, briefly.
5325  * @ingroup Parser
5326  */
5327 class StripState {
5328         var $general, $nowiki;
5329
5330         function __construct() {
5331                 $this->general = new ReplacementArray;
5332                 $this->nowiki = new ReplacementArray;
5333         }
5334
5335         function unstripGeneral( $text ) {
5336                 wfProfileIn( __METHOD__ );
5337                 do {
5338                         $oldText = $text;
5339                         $text = $this->general->replace( $text );
5340                 } while ( $text !== $oldText );
5341                 wfProfileOut( __METHOD__ );
5342                 return $text;
5343         }
5344
5345         function unstripNoWiki( $text ) {
5346                 wfProfileIn( __METHOD__ );
5347                 do {
5348                         $oldText = $text;
5349                         $text = $this->nowiki->replace( $text );
5350                 } while ( $text !== $oldText );
5351                 wfProfileOut( __METHOD__ );
5352                 return $text;
5353         }
5354
5355         function unstripBoth( $text ) {
5356                 wfProfileIn( __METHOD__ );
5357                 do {
5358                         $oldText = $text;
5359                         $text = $this->general->replace( $text );
5360                         $text = $this->nowiki->replace( $text );
5361                 } while ( $text !== $oldText );
5362                 wfProfileOut( __METHOD__ );
5363                 return $text;
5364         }
5365 }
5366
5367 /**
5368  * @todo document, briefly.
5369  * @ingroup Parser
5370  */
5371 class OnlyIncludeReplacer {
5372         var $output = '';
5373
5374         function replace( $matches ) {
5375                 if ( substr( $matches[1], -1 ) === "\n" ) {
5376                         $this->output .= substr( $matches[1], 0, -1 );
5377                 } else {
5378                         $this->output .= $matches[1];
5379                 }
5380         }
5381 }