includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; // empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; // number of expensive parser function calls
 106         var $mFileCache;
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 131                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 132                 $this->mUrlProtocols = wfUrlProtocols();
 133                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 134                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 135                 $this->mVarCache = array();
 136                 if ( isset( $conf['preprocessorClass'] ) ) {
 137                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 138                 } elseif ( extension_loaded( 'domxml' ) ) {
 139                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 140                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 141                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 142                 } elseif ( extension_loaded( 'dom' ) ) {
 143                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 144                 } else {
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 }
 147                 $this->mMarkerIndex = 0;
 148                 $this->mFirstCall = true;
 149         }
 150
 151         /**
 152          * Reduce memory usage to reduce the impact of circular references
 153          */
 154         function __destruct() {
 155                 if ( isset( $this->mLinkHolders ) ) {
 156                         $this->mLinkHolders->__destruct();
 157                 }
 158                 foreach ( $this as $name => $value ) {
 159                         unset( $this->$name );
 160                 }
 161         }
 162
 163         /**
 164          * Do various kinds of initialisation on the first call of the parser
 165          */
 166         function firstCallInit() {
 167                 if ( !$this->mFirstCall ) {
 168                         return;
 169                 }
 170                 $this->mFirstCall = false;
 171
 172                 wfProfileIn( __METHOD__ );
 173
 174                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 175                 CoreParserFunctions::register( $this );
 176                 $this->initialiseVariables();
 177
 178                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = new StripState;
 198                 $this->mArgStack = false;
 199                 $this->mInPre = false;
 200                 $this->mLinkHolders = new LinkHolderArray( $this );
 201                 $this->mLinkID = 0;
 202                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 203
 204                 /**
 205                  * Prefix for temporary replacement strings for the multipass parser.
 206                  * \x07 should never appear in input as it's disallowed in XML.
 207                  * Using it at the front also gives us a little extra robustness
 208                  * since it shouldn't match when butted up against identifier-like
 209                  * string constructs.
 210                  *
 211                  * Must not consist of all title characters, or else it will change
 212                  * the behaviour of <nowiki> in a link.
 213                  */
 214                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 215                 # Changed to \x7f to allow XML double-parsing -- TS
 216                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 217
 218
 219                 # Clear these on every parse, bug 4549
 220                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 221
 222                 $this->mShowToc = true;
 223                 $this->mForceTocPosition = false;
 224                 $this->mIncludeSizes = array(
 225                         'post-expand' => 0,
 226                         'arg' => 0,
 227                 );
 228                 $this->mPPNodeCount = 0;
 229                 $this->mDefaultSort = false;
 230                 $this->mHeadings = array();
 231                 $this->mDoubleUnderscores = array();
 232                 $this->mExpensiveFunctionCount = 0;
 233                 $this->mFileCache = array();
 234
 235                 # Fix cloning
 236                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 237                         $this->mPreprocessor = null;
 238                 }
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == self::OT_HTML,
 249                         'wiki' => $ot == self::OT_WIKI,
 250                         'pre' => $ot == self::OT_PREPROCESS,
 251                 );
 252         }
 253
 254         /**
 255          * Set the context title
 256          */
 257         function setTitle( $t ) {
 258                 if ( !$t || $t instanceof FakeTitle ) {
 259                         $t = Title::newFromText( 'NO TITLE' );
 260                 }
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if( !isset( $this->mUniqPrefix ) ) {
 277                         // @fixme this is probably *horribly wrong*
 278                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         // if this is called for a parser cache hit, the parser may not
 280                         // have ever been initialized in the first place.
 281                         // Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title );
 317                 $oldRevisionId = $this->mRevisionId;
 318                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 319                 if( $revid !== null ) {
 320                         $this->mRevisionId = $revid;
 321                         $this->mRevisionTimestamp = null;
 322                 }
 323                 $this->setOutputType( self::OT_HTML );
 324                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 # No more strip!
 326                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 $text = $this->internalParse( $text );
 328                 $text = $this->mStripState->unstripGeneral( $text );
 329
 330                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 331                 $fixtags = array(
 332                         # french spaces, last one Guillemet-left
 333                         # only if there is something before the space
 334                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 335                         # french spaces, Guillemet-right
 336                         '/(\\302\\253) /' => '\\1&nbsp;',
 337                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 338                 );
 339                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 340
 341                 $text = $this->doBlockLevels( $text, $linestart );
 342
 343                 $this->replaceLinkHolders( $text );
 344
 345                 # the position of the parserConvert() call should not be changed. it
 346                 # assumes that the links are all replaced and the only thing left
 347                 # is the <nowiki> mark.
 348                 # Side-effects: this calls $this->mOutput->setTitleText()
 349                 $text = $wgContLang->parserConvert( $text, $this );
 350
 351                 $text = $this->mStripState->unstripNoWiki( $text );
 352
 353                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 354
 355 //!JF Move to its own function
 356
 357                 $uniq_prefix = $this->mUniqPrefix;
 358                 $matches = array();
 359                 $elements = array_keys( $this->mTransparentTagHooks );
 360                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 361
 362                 foreach( $matches as $marker => $data ) {
 363                         list( $element, $content, $params, $tag ) = $data;
 364                         $tagName = strtolower( $element );
 365                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 366                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 367                                         array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 378                         $text = MWTidy::tidy( $text );
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         global $wgExpensiveParserFunctionLimit;
 416                         $max = $this->mOptions->getMaxIncludeSize();
 417                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 418                         $limitReport =
 419                                 "NewPP limit report\n" .
 420                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 421                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 422                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 423                                 $PFreport;
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          */
 440         function recursiveTagParse( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 444                 $text = $this->internalParse( $text );
 445                 wfProfileOut( __METHOD__ );
 446                 return $text;
 447         }
 448
 449         /**
 450          * Expand templates and variables in the text, producing valid, static wikitext.
 451          * Also removes comments.
 452          */
 453         function preprocess( $text, $title, $options, $revid = null ) {
 454                 wfProfileIn( __METHOD__ );
 455                 $this->clearState();
 456                 $this->setOutputType( self::OT_PREPROCESS );
 457                 $this->mOptions = $options;
 458                 $this->setTitle( $title );
 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          * Get a random string
 472          *
 473          * @private
 474          * @static
 475          */
 476         function getRandomString() {
 477                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 478         }
 479
 480         function &getTitle() { return $this->mTitle; }
 481         function getOptions() { return $this->mOptions; }
 482         function getRevisionId() { return $this->mRevisionId; }
 483         function getOutput() { return $this->mOutput; }
 484         function nextLinkID() { return $this->mLinkID++; }
 485
 486         function getFunctionLang() {
 487                 global $wgLang, $wgContLang;
 488
 489                 $target = $this->mOptions->getTargetLanguage();
 490                 if ( $target !== null ) {
 491                         return $target;
 492                 } else {
 493                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 494                 }
 495         }
 496
 497         /**
 498          * Get a preprocessor object
 499          */
 500         function getPreprocessor() {
 501                 if ( !isset( $this->mPreprocessor ) ) {
 502                         $class = $this->mPreprocessorClass;
 503                         $this->mPreprocessor = new $class( $this );
 504                 }
 505                 return $this->mPreprocessor;
 506         }
 507
 508         /**
 509          * Replaces all occurrences of HTML-style comments and the given tags
 510          * in the text with a random marker and returns the next text. The output
 511          * parameter $matches will be an associative array filled with data in
 512          * the form:
 513          *   'UNIQ-xxxxx' => array(
 514          *     'element',
 515          *     'tag content',
 516          *     array( 'param' => 'x' ),
 517          *     '<element param="x">tag content</element>' ) )
 518          *
 519          * @param $elements list of element names. Comments are always extracted.
 520          * @param $text Source text string.
 521          * @param $uniq_prefix
 522          *
 523          * @public
 524          * @static
 525          */
 526         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 527                 static $n = 1;
 528                 $stripped = '';
 529                 $matches = array();
 530
 531                 $taglist = implode( '|', $elements );
 532                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 533
 534                 while ( '' != $text ) {
 535                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 536                         $stripped .= $p[0];
 537                         if( count( $p ) < 5 ) {
 538                                 break;
 539                         }
 540                         if( count( $p ) > 5 ) {
 541                                 // comment
 542                                 $element    = $p[4];
 543                                 $attributes = '';
 544                                 $close      = '';
 545                                 $inside     = $p[5];
 546                         } else {
 547                                 // tag
 548                                 $element    = $p[1];
 549                                 $attributes = $p[2];
 550                                 $close      = $p[3];
 551                                 $inside     = $p[4];
 552                         }
 553
 554                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 555                         $stripped .= $marker;
 556
 557                         if ( $close === '/>' ) {
 558                                 // Empty element tag, <tag />
 559                                 $content = null;
 560                                 $text = $inside;
 561                                 $tail = null;
 562                         } else {
 563                                 if( $element === '!--' ) {
 564                                         $end = '/(-->)/';
 565                                 } else {
 566                                         $end = "/(<\\/$element\\s*>)/i";
 567                                 }
 568                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 569                                 $content = $q[0];
 570                                 if( count( $q ) < 3 ) {
 571                                         # No end tag -- let it run out to the end of the text.
 572                                         $tail = '';
 573                                         $text = '';
 574                                 } else {
 575                                         $tail = $q[1];
 576                                         $text = $q[2];
 577                                 }
 578                         }
 579
 580                         $matches[$marker] = array( $element,
 581                                 $content,
 582                                 Sanitizer::decodeTagAttributes( $attributes ),
 583                                 "<$element$attributes$close$content$tail" );
 584                 }
 585                 return $stripped;
 586         }
 587
 588         /**
 589          * Get a list of strippable XML-like elements
 590          */
 591         function getStripList() {
 592                 global $wgRawHtml;
 593                 $elements = $this->mStripList;
 594                 if( $wgRawHtml ) {
 595                         $elements[] = 'html';
 596                 }
 597                 if( $this->mOptions->getUseTeX() ) {
 598                         $elements[] = 'math';
 599                 }
 600                 return $elements;
 601         }
 602
 603         /**
 604          * @deprecated use replaceVariables
 605          */
 606         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 607                 return $text;
 608         }
 609
 610         /**
 611          * Restores pre, math, and other extensions removed by strip()
 612          *
 613          * always call unstripNoWiki() after this one
 614          * @private
 615          * @deprecated use $this->mStripState->unstrip()
 616          */
 617         function unstrip( $text, $state ) {
 618                 return $state->unstripGeneral( $text );
 619         }
 620
 621         /**
 622          * Always call this after unstrip() to preserve the order
 623          *
 624          * @private
 625          * @deprecated use $this->mStripState->unstrip()
 626          */
 627         function unstripNoWiki( $text, $state ) {
 628                 return $state->unstripNoWiki( $text );
 629         }
 630
 631         /**
 632          * @deprecated use $this->mStripState->unstripBoth()
 633          */
 634         function unstripForHTML( $text ) {
 635                 return $this->mStripState->unstripBoth( $text );
 636         }
 637
 638         /**
 639          * Add an item to the strip state
 640          * Returns the unique tag which must be inserted into the stripped text
 641          * The tag will be replaced with the original text in unstrip()
 642          *
 643          * @private
 644          */
 645         function insertStripItem( $text ) {
 646                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 647                 $this->mMarkerIndex++;
 648                 $this->mStripState->general->setPair( $rnd, $text );
 649                 return $rnd;
 650         }
 651
 652         /**
 653          * Interface with html tidy
 654          * @deprecated Use MWTidy::tidy()
 655          */
 656         public static function tidy( $text ) {
 657                 wfDeprecated( __METHOD__ );
 658                 return MWTidy::tidy( $text );
 659         }
 660
 661         /**
 662          * parse the wiki syntax used to render tables
 663          *
 664          * @private
 665          */
 666         function doTableStuff ( $text ) {
 667                 wfProfileIn( __METHOD__ );
 668
 669                 $lines = StringUtils::explode( "\n", $text );
 670                 $out = '';
 671                 $td_history = array (); // Is currently a td tag open?
 672                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 673                 $tr_history = array (); // Is currently a tr tag open?
 674                 $tr_attributes = array (); // history of tr attributes
 675                 $has_opened_tr = array(); // Did this table open a <tr> element?
 676                 $indent_level = 0; // indent level of the table
 677
 678                 foreach ( $lines as $outLine ) {
 679                         $line = trim( $outLine );
 680
 681                         if( $line == '' ) { // empty line, go to next line
 682                                 $out .= $outLine."\n";
 683                                 continue;
 684                         }
 685                         $first_character = $line[0];
 686                         $matches = array();
 687
 688                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 689                                 // First check if we are starting a new table
 690                                 $indent_level = strlen( $matches[1] );
 691
 692                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 693                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 694
 695                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 696                                 array_push ( $td_history , false );
 697                                 array_push ( $last_tag_history , '' );
 698                                 array_push ( $tr_history , false );
 699                                 array_push ( $tr_attributes , '' );
 700                                 array_push ( $has_opened_tr , false );
 701                         } else if ( count ( $td_history ) == 0 ) {
 702                                 // Don't do any of the following
 703                                 $out .= $outLine."\n";
 704                                 continue;
 705                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 706                                 // We are ending a table
 707                                 $line = '</table>' . substr ( $line , 2 );
 708                                 $last_tag = array_pop ( $last_tag_history );
 709
 710                                 if ( !array_pop ( $has_opened_tr ) ) {
 711                                         $line = "<tr><td></td></tr>{$line}";
 712                                 }
 713
 714                                 if ( array_pop ( $tr_history ) ) {
 715                                         $line = "</tr>{$line}";
 716                                 }
 717
 718                                 if ( array_pop ( $td_history ) ) {
 719                                         $line = "</{$last_tag}>{$line}";
 720                                 }
 721                                 array_pop ( $tr_attributes );
 722                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 723                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 724                                 // Now we have a table row
 725                                 $line = preg_replace( '#^\|-+#', '', $line );
 726
 727                                 // Whats after the tag is now only attributes
 728                                 $attributes = $this->mStripState->unstripBoth( $line );
 729                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 730                                 array_pop ( $tr_attributes );
 731                                 array_push ( $tr_attributes , $attributes );
 732
 733                                 $line = '';
 734                                 $last_tag = array_pop ( $last_tag_history );
 735                                 array_pop ( $has_opened_tr );
 736                                 array_push ( $has_opened_tr , true );
 737
 738                                 if ( array_pop ( $tr_history ) ) {
 739                                         $line = '</tr>';
 740                                 }
 741
 742                                 if ( array_pop ( $td_history ) ) {
 743                                         $line = "</{$last_tag}>{$line}";
 744                                 }
 745
 746                                 $outLine = $line;
 747                                 array_push ( $tr_history , false );
 748                                 array_push ( $td_history , false );
 749                                 array_push ( $last_tag_history , '' );
 750                         }
 751                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 752                                 // This might be cell elements, td, th or captions
 753                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 754                                         $first_character = '+';
 755                                         $line = substr ( $line , 1 );
 756                                 }
 757
 758                                 $line = substr ( $line , 1 );
 759
 760                                 if ( $first_character === '!' ) {
 761                                         $line = str_replace ( '!!' , '||' , $line );
 762                                 }
 763
 764                                 // Split up multiple cells on the same line.
 765                                 // FIXME : This can result in improper nesting of tags processed
 766                                 // by earlier parser steps, but should avoid splitting up eg
 767                                 // attribute values containing literal "||".
 768                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 769
 770                                 $outLine = '';
 771
 772                                 // Loop through each table cell
 773                                 foreach ( $cells as $cell )
 774                                 {
 775                                         $previous = '';
 776                                         if ( $first_character !== '+' )
 777                                         {
 778                                                 $tr_after = array_pop ( $tr_attributes );
 779                                                 if ( !array_pop ( $tr_history ) ) {
 780                                                         $previous = "<tr{$tr_after}>\n";
 781                                                 }
 782                                                 array_push ( $tr_history , true );
 783                                                 array_push ( $tr_attributes , '' );
 784                                                 array_pop ( $has_opened_tr );
 785                                                 array_push ( $has_opened_tr , true );
 786                                         }
 787
 788                                         $last_tag = array_pop ( $last_tag_history );
 789
 790                                         if ( array_pop ( $td_history ) ) {
 791                                                 $previous = "</{$last_tag}>{$previous}";
 792                                         }
 793
 794                                         if ( $first_character === '|' ) {
 795                                                 $last_tag = 'td';
 796                                         } else if ( $first_character === '!' ) {
 797                                                 $last_tag = 'th';
 798                                         } else if ( $first_character === '+' ) {
 799                                                 $last_tag = 'caption';
 800                                         } else {
 801                                                 $last_tag = '';
 802                                         }
 803
 804                                         array_push ( $last_tag_history , $last_tag );
 805
 806                                         // A cell could contain both parameters and data
 807                                         $cell_data = explode ( '|' , $cell , 2 );
 808
 809                                         // Bug 553: Note that a '|' inside an invalid link should not
 810                                         // be mistaken as delimiting cell parameters
 811                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 812                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 813                                         } else if ( count ( $cell_data ) == 1 )
 814                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 815                                         else {
 816                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 817                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 818                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 819                                         }
 820
 821                                         $outLine .= $cell;
 822                                         array_push ( $td_history , true );
 823                                 }
 824                         }
 825                         $out .= $outLine . "\n";
 826                 }
 827
 828                 // Closing open td, tr && table
 829                 while ( count ( $td_history ) > 0 )
 830                 {
 831                         if ( array_pop ( $td_history ) ) {
 832                                 $out .= "</td>\n";
 833                         }
 834                         if ( array_pop ( $tr_history ) ) {
 835                                 $out .= "</tr>\n";
 836                         }
 837                         if ( !array_pop ( $has_opened_tr ) ) {
 838                                 $out .= "<tr><td></td></tr>\n" ;
 839                         }
 840
 841                         $out .= "</table>\n";
 842                 }
 843
 844                 // Remove trailing line-ending (b/c)
 845                 if ( substr( $out, -1 ) === "\n" ) {
 846                         $out = substr( $out, 0, -1 );
 847                 }
 848
 849                 // special case: don't return empty table
 850                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 851                         $out = '';
 852                 }
 853
 854                 wfProfileOut( __METHOD__ );
 855
 856                 return $out;
 857         }
 858
 859         /**
 860          * Helper function for parse() that transforms wiki markup into
 861          * HTML. Only called for $mOutputType == self::OT_HTML.
 862          *
 863          * @private
 864          */
 865         function internalParse( $text ) {
 866                 $isMain = true;
 867                 wfProfileIn( __METHOD__ );
 868
 869                 $origText = $text;
 870
 871                 # Hook to suspend the parser in this state
 872                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 873                         wfProfileOut( __METHOD__ );
 874                         return $text ;
 875                 }
 876
 877                 $text = $this->replaceVariables( $text );
 878                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 879                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 880
 881                 // Tables need to come after variable replacement for things to work
 882                 // properly; putting them before other transformations should keep
 883                 // exciting things like link expansions from showing up in surprising
 884                 // places.
 885                 $text = $this->doTableStuff( $text );
 886
 887                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 888
 889                 $text = $this->doDoubleUnderscore( $text );
 890                 $text = $this->doHeadings( $text );
 891                 if( $this->mOptions->getUseDynamicDates() ) {
 892                         $df = DateFormatter::getInstance();
 893                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 894                 }
 895                 $text = $this->doAllQuotes( $text );
 896                 $text = $this->replaceInternalLinks( $text );
 897                 $text = $this->replaceExternalLinks( $text );
 898
 899                 # replaceInternalLinks may sometimes leave behind
 900                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 901                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 902
 903                 $text = $this->doMagicLinks( $text );
 904                 $text = $this->formatHeadings( $text, $origText, $isMain );
 905
 906                 wfProfileOut( __METHOD__ );
 907                 return $text;
 908         }
 909
 910         /**
 911          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 912          * magic external links.
 913          *
 914          * DML
 915          * @private
 916          */
 917         function doMagicLinks( $text ) {
 918                 wfProfileIn( __METHOD__ );
 919                 $prots = $this->mUrlProtocols;
 920                 $urlChar = self::EXT_LINK_URL_CLASS;
 921                 $text = preg_replace_callback(
 922                         '!(?:                           # Start cases
 923                                 (<a.*?</a>) |               # m[1]: Skip link text
 924                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 925                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 926                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 927                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 928                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 929                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 930                                     [0-9Xx]                 # check digit
 931                                     \b)
 932                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 933                 wfProfileOut( __METHOD__ );
 934                 return $text;
 935         }
 936
 937         function magicLinkCallback( $m ) {
 938                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 939                         # Skip anchor
 940                         return $m[0];
 941                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 942                         # Skip HTML element
 943                         return $m[0];
 944                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 945                         # Free external link
 946                         return $this->makeFreeExternalLink( $m[0] );
 947                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 948                         # RFC or PMID
 949                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 950                                 $keyword = 'RFC';
 951                                 $urlmsg = 'rfcurl';
 952                                 $id = $m[4];
 953                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 954                                 $keyword = 'PMID';
 955                                 $urlmsg = 'pubmedurl';
 956                                 $id = $m[4];
 957                         } else {
 958                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 959                                         substr($m[0], 0, 20 ) . '"' );
 960                         }
 961                         $url = wfMsg( $urlmsg, $id);
 962                         $sk = $this->mOptions->getSkin();
 963                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
 964                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 965                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
 966                         # ISBN
 967                         $isbn = $m[5];
 968                         $num = strtr( $isbn, array(
 969                                 '-' => '',
 970                                 ' ' => '',
 971                                 'x' => 'X',
 972                         ));
 973                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
 974                         return'<a href="' .
 975                                 $titleObj->escapeLocalUrl() .
 976                                 "\" class=\"internal\">ISBN $isbn</a>";
 977                 } else {
 978                         return $m[0];
 979                 }
 980         }
 981
 982         /**
 983          * Make a free external link, given a user-supplied URL
 984          * @return HTML
 985          * @private
 986          */
 987         function makeFreeExternalLink( $url ) {
 988                 global $wgContLang;
 989                 wfProfileIn( __METHOD__ );
 990
 991                 $sk = $this->mOptions->getSkin();
 992                 $trail = '';
 993
 994                 # The characters '<' and '>' (which were escaped by
 995                 # removeHTMLtags()) should not be included in
 996                 # URLs, per RFC 2396.
 997                 $m2 = array();
 998                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
 999                         $trail = substr($url, $m2[0][1]) . $trail;
1000                         $url = substr($url, 0, $m2[0][1]);
1001                 }
1002
1003                 # Move trailing punctuation to $trail
1004                 $sep = ',;\.:!?';
1005                 # If there is no left bracket, then consider right brackets fair game too
1006                 if ( strpos( $url, '(' ) === false ) {
1007                         $sep .= ')';
1008                 }
1009
1010                 $numSepChars = strspn( strrev( $url ), $sep );
1011                 if ( $numSepChars ) {
1012                         $trail = substr( $url, -$numSepChars ) . $trail;
1013                         $url = substr( $url, 0, -$numSepChars );
1014                 }
1015
1016                 $url = Sanitizer::cleanUrl( $url );
1017
1018                 # Is this an external image?
1019                 $text = $this->maybeMakeExternalImage( $url );
1020                 if ( $text === false ) {
1021                         # Not an image, make a link
1022                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1023                                 $this->getExternalLinkAttribs( $url ) );
1024                         # Register it in the output object...
1025                         # Replace unnecessary URL escape codes with their equivalent characters
1026                         $pasteurized = self::replaceUnusualEscapes( $url );
1027                         $this->mOutput->addExternalLink( $pasteurized );
1028                 }
1029                 wfProfileOut( __METHOD__ );
1030                 return $text . $trail;
1031         }
1032
1033
1034         /**
1035          * Parse headers and return html
1036          *
1037          * @private
1038          */
1039         function doHeadings( $text ) {
1040                 wfProfileIn( __METHOD__ );
1041                 for ( $i = 6; $i >= 1; --$i ) {
1042                         $h = str_repeat( '=', $i );
1043                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1044                           "<h$i>\\1</h$i>", $text );
1045                 }
1046                 wfProfileOut( __METHOD__ );
1047                 return $text;
1048         }
1049
1050         /**
1051          * Replace single quotes with HTML markup
1052          * @private
1053          * @return string the altered text
1054          */
1055         function doAllQuotes( $text ) {
1056                 wfProfileIn( __METHOD__ );
1057                 $outtext = '';
1058                 $lines = StringUtils::explode( "\n", $text );
1059                 foreach ( $lines as $line ) {
1060                         $outtext .= $this->doQuotes( $line ) . "\n";
1061                 }
1062                 $outtext = substr($outtext, 0,-1);
1063                 wfProfileOut( __METHOD__ );
1064                 return $outtext;
1065         }
1066
1067         /**
1068          * Helper function for doAllQuotes()
1069          */
1070         public function doQuotes( $text ) {
1071                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1072                 if ( count( $arr ) == 1 )
1073                         return $text;
1074                 else
1075                 {
1076                         # First, do some preliminary work. This may shift some apostrophes from
1077                         # being mark-up to being text. It also counts the number of occurrences
1078                         # of bold and italics mark-ups.
1079                         $i = 0;
1080                         $numbold = 0;
1081                         $numitalics = 0;
1082                         foreach ( $arr as $r )
1083                         {
1084                                 if ( ( $i % 2 ) == 1 )
1085                                 {
1086                                         # If there are ever four apostrophes, assume the first is supposed to
1087                                         # be text, and the remaining three constitute mark-up for bold text.
1088                                         if ( strlen( $arr[$i] ) == 4 )
1089                                         {
1090                                                 $arr[$i-1] .= "'";
1091                                                 $arr[$i] = "'''";
1092                                         }
1093                                         # If there are more than 5 apostrophes in a row, assume they're all
1094                                         # text except for the last 5.
1095                                         else if ( strlen( $arr[$i] ) > 5 )
1096                                         {
1097                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1098                                                 $arr[$i] = "'''''";
1099                                         }
1100                                         # Count the number of occurrences of bold and italics mark-ups.
1101                                         # We are not counting sequences of five apostrophes.
1102                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1103                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1104                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1105                                 }
1106                                 $i++;
1107                         }
1108
1109                         # If there is an odd number of both bold and italics, it is likely
1110                         # that one of the bold ones was meant to be an apostrophe followed
1111                         # by italics. Which one we cannot know for certain, but it is more
1112                         # likely to be one that has a single-letter word before it.
1113                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1114                         {
1115                                 $i = 0;
1116                                 $firstsingleletterword = -1;
1117                                 $firstmultiletterword = -1;
1118                                 $firstspace = -1;
1119                                 foreach ( $arr as $r )
1120                                 {
1121                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1122                                         {
1123                                                 $x1 = substr ($arr[$i-1], -1);
1124                                                 $x2 = substr ($arr[$i-1], -2, 1);
1125                                                 if ($x1 === ' ') {
1126                                                         if ($firstspace == -1) $firstspace = $i;
1127                                                 } else if ($x2 === ' ') {
1128                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1129                                                 } else {
1130                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1131                                                 }
1132                                         }
1133                                         $i++;
1134                                 }
1135
1136                                 # If there is a single-letter word, use it!
1137                                 if ($firstsingleletterword > -1)
1138                                 {
1139                                         $arr [ $firstsingleletterword ] = "''";
1140                                         $arr [ $firstsingleletterword-1 ] .= "'";
1141                                 }
1142                                 # If not, but there's a multi-letter word, use that one.
1143                                 else if ($firstmultiletterword > -1)
1144                                 {
1145                                         $arr [ $firstmultiletterword ] = "''";
1146                                         $arr [ $firstmultiletterword-1 ] .= "'";
1147                                 }
1148                                 # ... otherwise use the first one that has neither.
1149                                 # (notice that it is possible for all three to be -1 if, for example,
1150                                 # there is only one pentuple-apostrophe in the line)
1151                                 else if ($firstspace > -1)
1152                                 {
1153                                         $arr [ $firstspace ] = "''";
1154                                         $arr [ $firstspace-1 ] .= "'";
1155                                 }
1156                         }
1157
1158                         # Now let's actually convert our apostrophic mush to HTML!
1159                         $output = '';
1160                         $buffer = '';
1161                         $state = '';
1162                         $i = 0;
1163                         foreach ($arr as $r)
1164                         {
1165                                 if (($i % 2) == 0)
1166                                 {
1167                                         if ($state === 'both')
1168                                                 $buffer .= $r;
1169                                         else
1170                                                 $output .= $r;
1171                                 }
1172                                 else
1173                                 {
1174                                         if (strlen ($r) == 2)
1175                                         {
1176                                                 if ($state === 'i')
1177                                                 { $output .= '</i>'; $state = ''; }
1178                                                 else if ($state === 'bi')
1179                                                 { $output .= '</i>'; $state = 'b'; }
1180                                                 else if ($state === 'ib')
1181                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1182                                                 else if ($state === 'both')
1183                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1184                                                 else # $state can be 'b' or ''
1185                                                 { $output .= '<i>'; $state .= 'i'; }
1186                                         }
1187                                         else if (strlen ($r) == 3)
1188                                         {
1189                                                 if ($state === 'b')
1190                                                 { $output .= '</b>'; $state = ''; }
1191                                                 else if ($state === 'bi')
1192                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1193                                                 else if ($state === 'ib')
1194                                                 { $output .= '</b>'; $state = 'i'; }
1195                                                 else if ($state === 'both')
1196                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1197                                                 else # $state can be 'i' or ''
1198                                                 { $output .= '<b>'; $state .= 'b'; }
1199                                         }
1200                                         else if (strlen ($r) == 5)
1201                                         {
1202                                                 if ($state === 'b')
1203                                                 { $output .= '</b><i>'; $state = 'i'; }
1204                                                 else if ($state === 'i')
1205                                                 { $output .= '</i><b>'; $state = 'b'; }
1206                                                 else if ($state === 'bi')
1207                                                 { $output .= '</i></b>'; $state = ''; }
1208                                                 else if ($state === 'ib')
1209                                                 { $output .= '</b></i>'; $state = ''; }
1210                                                 else if ($state === 'both')
1211                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1212                                                 else # ($state == '')
1213                                                 { $buffer = ''; $state = 'both'; }
1214                                         }
1215                                 }
1216                                 $i++;
1217                         }
1218                         # Now close all remaining tags.  Notice that the order is important.
1219                         if ($state === 'b' || $state === 'ib')
1220                                 $output .= '</b>';
1221                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1222                                 $output .= '</i>';
1223                         if ($state === 'bi')
1224                                 $output .= '</b>';
1225                         # There might be lonely ''''', so make sure we have a buffer
1226                         if ($state === 'both' && $buffer)
1227                                 $output .= '<b><i>'.$buffer.'</i></b>';
1228                         return $output;
1229                 }
1230         }
1231
1232         /**
1233          * Replace external links (REL)
1234          *
1235          * Note: this is all very hackish and the order of execution matters a lot.
1236          * Make sure to run maintenance/parserTests.php if you change this code.
1237          *
1238          * @private
1239          */
1240         function replaceExternalLinks( $text ) {
1241                 global $wgContLang;
1242                 wfProfileIn( __METHOD__ );
1243
1244                 $sk = $this->mOptions->getSkin();
1245
1246                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1247                 $s = array_shift( $bits );
1248
1249                 $i = 0;
1250                 while ( $i<count( $bits ) ) {
1251                         $url = $bits[$i++];
1252                         $protocol = $bits[$i++];
1253                         $text = $bits[$i++];
1254                         $trail = $bits[$i++];
1255
1256                         # The characters '<' and '>' (which were escaped by
1257                         # removeHTMLtags()) should not be included in
1258                         # URLs, per RFC 2396.
1259                         $m2 = array();
1260                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1261                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1262                                 $url = substr($url, 0, $m2[0][1]);
1263                         }
1264
1265                         # If the link text is an image URL, replace it with an <img> tag
1266                         # This happened by accident in the original parser, but some people used it extensively
1267                         $img = $this->maybeMakeExternalImage( $text );
1268                         if ( $img !== false ) {
1269                                 $text = $img;
1270                         }
1271
1272                         $dtrail = '';
1273
1274                         # Set linktype for CSS - if URL==text, link is essentially free
1275                         $linktype = ($text === $url) ? 'free' : 'text';
1276
1277                         # No link text, e.g. [http://domain.tld/some.link]
1278                         if ( $text == '' ) {
1279                                 # Autonumber if allowed. See bug #5918
1280                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1281                                         $langObj = $this->getFunctionLang();
1282                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1283                                         $linktype = 'autonumber';
1284                                 } else {
1285                                         # Otherwise just use the URL
1286                                         $text = htmlspecialchars( $url );
1287                                         $linktype = 'free';
1288                                 }
1289                         } else {
1290                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1291                                 # Check for trail
1292                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1293                         }
1294
1295                         $text = $wgContLang->markNoConversion($text);
1296
1297                         $url = Sanitizer::cleanUrl( $url );
1298
1299                         # Use the encoded URL
1300                         # This means that users can paste URLs directly into the text
1301                         # Funny characters like &ouml; aren't valid in URLs anyway
1302                         # This was changed in August 2004
1303                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1304                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1305
1306                         # Register link in the output object.
1307                         # Replace unnecessary URL escape codes with the referenced character
1308                         # This prevents spammers from hiding links from the filters
1309                         $pasteurized = self::replaceUnusualEscapes( $url );
1310                         $this->mOutput->addExternalLink( $pasteurized );
1311                 }
1312
1313                 wfProfileOut( __METHOD__ );
1314                 return $s;
1315         }
1316
1317         /**
1318          * Get an associative array of additional HTML attributes appropriate for a
1319          * particular external link.  This currently may include rel => nofollow
1320          * (depending on configuration, namespace, and the URL's domain) and/or a
1321          * target attribute (depending on configuration).
1322          *
1323          * @param string $url Optional URL, to extract the domain from for rel =>
1324          *   nofollow if appropriate
1325          * @return array Associative array of HTML attributes
1326          */
1327         function getExternalLinkAttribs( $url = false ) {
1328                 $attribs = array();
1329                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1330                 $ns = $this->mTitle->getNamespace();
1331                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1332                         $attribs['rel'] = 'nofollow';
1333
1334                         global $wgNoFollowDomainExceptions;
1335                         if ( $wgNoFollowDomainExceptions ) {
1336                                 $bits = wfParseUrl( $url );
1337                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1338                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1339                                                 if( substr( $bits['host'], -strlen( $domain ) )
1340                                                 == $domain ) {
1341                                                         unset( $attribs['rel'] );
1342                                                         break;
1343                                                 }
1344                                         }
1345                                 }
1346                         }
1347                 }
1348                 if ( $this->mOptions->getExternalLinkTarget() ) {
1349                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1350                 }
1351                 return $attribs;
1352         }
1353
1354
1355         /**
1356          * Replace unusual URL escape codes with their equivalent characters
1357          * @param string
1358          * @return string
1359          * @static
1360          * @todo  This can merge genuinely required bits in the path or query string,
1361          *        breaking legit URLs. A proper fix would treat the various parts of
1362          *        the URL differently; as a workaround, just use the output for
1363          *        statistical records, not for actual linking/output.
1364          */
1365         static function replaceUnusualEscapes( $url ) {
1366                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1367                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1368         }
1369
1370         /**
1371          * Callback function used in replaceUnusualEscapes().
1372          * Replaces unusual URL escape codes with their equivalent character
1373          * @static
1374          * @private
1375          */
1376         private static function replaceUnusualEscapesCallback( $matches ) {
1377                 $char = urldecode( $matches[0] );
1378                 $ord = ord( $char );
1379                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1380                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1381                         // No, shouldn't be escaped
1382                         return $char;
1383                 } else {
1384                         // Yes, leave it escaped
1385                         return $matches[0];
1386                 }
1387         }
1388
1389         /**
1390          * make an image if it's allowed, either through the global
1391          * option, through the exception, or through the on-wiki whitelist
1392          * @private
1393          */
1394         function maybeMakeExternalImage( $url ) {
1395                 $sk = $this->mOptions->getSkin();
1396                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1397                 $imagesexception = !empty($imagesfrom);
1398                 $text = false;
1399                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1400                 if( $imagesexception && is_array( $imagesfrom ) ) {
1401                         $imagematch = false;
1402                         foreach( $imagesfrom as $match ) {
1403                                 if( strpos( $url, $match ) === 0 ) {
1404                                         $imagematch = true;
1405                                         break;
1406                                 }
1407                         }
1408                 } elseif( $imagesexception ) {
1409                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1410                 } else {
1411                         $imagematch = false;
1412                 }
1413                 if ( $this->mOptions->getAllowExternalImages()
1414                      || ( $imagesexception && $imagematch ) ) {
1415                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1416                                 # Image found
1417                                 $text = $sk->makeExternalImage( $url );
1418                         }
1419                 }
1420                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1421                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1422                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1423                         foreach( $whitelist as $entry ) {
1424                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1425                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1426                                         continue;
1427                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1428                                         # Image matches a whitelist entry
1429                                         $text = $sk->makeExternalImage( $url );
1430                                         break;
1431                                 }
1432                         }
1433                 }
1434                 return $text;
1435         }
1436
1437         /**
1438          * Process [[ ]] wikilinks
1439          * @return processed text
1440          *
1441          * @private
1442          */
1443         function replaceInternalLinks( $s ) {
1444                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1445                 return $s;
1446         }
1447
1448         /**
1449          * Process [[ ]] wikilinks (RIL)
1450          * @return LinkHolderArray
1451          *
1452          * @private
1453          */
1454         function replaceInternalLinks2( &$s ) {
1455                 global $wgContLang;
1456
1457                 wfProfileIn( __METHOD__ );
1458
1459                 wfProfileIn( __METHOD__.'-setup' );
1460                 static $tc = FALSE, $e1, $e1_img;
1461                 # the % is needed to support urlencoded titles as well
1462                 if ( !$tc ) {
1463                         $tc = Title::legalChars() . '#%';
1464                         # Match a link having the form [[namespace:link|alternate]]trail
1465                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1466                         # Match cases where there is no "]]", which might still be images
1467                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1468                 }
1469
1470                 $sk = $this->mOptions->getSkin();
1471                 $holders = new LinkHolderArray( $this );
1472
1473                 #split the entire text string on occurences of [[
1474                 $a = StringUtils::explode( '[[', ' ' . $s );
1475                 #get the first element (all text up to first [[), and remove the space we added
1476                 $s = $a->current();
1477                 $a->next();
1478                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1479                 $s = substr( $s, 1 );
1480
1481                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1482                 $e2 = null;
1483                 if ( $useLinkPrefixExtension ) {
1484                         # Match the end of a line for a word that's not followed by whitespace,
1485                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1486                         $e2 = wfMsgForContent( 'linkprefix' );
1487                 }
1488
1489                 if( is_null( $this->mTitle ) ) {
1490                         wfProfileOut( __METHOD__.'-setup' );
1491                         wfProfileOut( __METHOD__ );
1492                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1493                 }
1494                 $nottalk = !$this->mTitle->isTalkPage();
1495
1496                 if ( $useLinkPrefixExtension ) {
1497                         $m = array();
1498                         if ( preg_match( $e2, $s, $m ) ) {
1499                                 $first_prefix = $m[2];
1500                         } else {
1501                                 $first_prefix = false;
1502                         }
1503                 } else {
1504                         $prefix = '';
1505                 }
1506
1507                 if($wgContLang->hasVariants()) {
1508                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1509                 } else {
1510                         $selflink = array($this->mTitle->getPrefixedText());
1511                 }
1512                 $useSubpages = $this->areSubpagesAllowed();
1513                 wfProfileOut( __METHOD__.'-setup' );
1514
1515                 # Loop for each link
1516                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1517                         # Check for excessive memory usage
1518                         if ( $holders->isBig() ) {
1519                                 # Too big
1520                                 # Do the existence check, replace the link holders and clear the array
1521                                 $holders->replace( $s );
1522                                 $holders->clear();
1523                         }
1524
1525                         if ( $useLinkPrefixExtension ) {
1526                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1527                                 if ( preg_match( $e2, $s, $m ) ) {
1528                                         $prefix = $m[2];
1529                                         $s = $m[1];
1530                                 } else {
1531                                         $prefix='';
1532                                 }
1533                                 # first link
1534                                 if($first_prefix) {
1535                                         $prefix = $first_prefix;
1536                                         $first_prefix = false;
1537                                 }
1538                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1539                         }
1540
1541                         $might_be_img = false;
1542
1543                         wfProfileIn( __METHOD__."-e1" );
1544                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1545                                 $text = $m[2];
1546                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1547                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1548                                 # the real problem is with the $e1 regex
1549                                 # See bug 1300.
1550                                 #
1551                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1552                                 # and no image is in sight. See bug 2095.
1553                                 #
1554                                 if( $text !== '' &&
1555                                         substr( $m[3], 0, 1 ) === ']' &&
1556                                         strpos($text, '[') !== false
1557                                 )
1558                                 {
1559                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1560                                         $m[3] = substr( $m[3], 1 );
1561                                 }
1562                                 # fix up urlencoded title texts
1563                                 if( strpos( $m[1], '%' ) !== false ) {
1564                                         # Should anchors '#' also be rejected?
1565                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1566                                 }
1567                                 $trail = $m[3];
1568                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1569                                 $might_be_img = true;
1570                                 $text = $m[2];
1571                                 if ( strpos( $m[1], '%' ) !== false ) {
1572                                         $m[1] = urldecode($m[1]);
1573                                 }
1574                                 $trail = "";
1575                         } else { # Invalid form; output directly
1576                                 $s .= $prefix . '[[' . $line ;
1577                                 wfProfileOut( __METHOD__."-e1" );
1578                                 continue;
1579                         }
1580                         wfProfileOut( __METHOD__."-e1" );
1581                         wfProfileIn( __METHOD__."-misc" );
1582
1583                         # Don't allow internal links to pages containing
1584                         # PROTO: where PROTO is a valid URL protocol; these
1585                         # should be external links.
1586                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1587                                 $s .= $prefix . '[[' . $line ;
1588                                 wfProfileOut( __METHOD__."-misc" );
1589                                 continue;
1590                         }
1591
1592                         # Make subpage if necessary
1593                         if( $useSubpages ) {
1594                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1595                         } else {
1596                                 $link = $m[1];
1597                         }
1598
1599                         $noforce = (substr($m[1], 0, 1) !== ':');
1600                         if (!$noforce) {
1601                                 # Strip off leading ':'
1602                                 $link = substr($link, 1);
1603                         }
1604
1605                         wfProfileOut( __METHOD__."-misc" );
1606                         wfProfileIn( __METHOD__."-title" );
1607                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1608                         if( $nt === NULL ) {
1609                                 $s .= $prefix . '[[' . $line;
1610                                 wfProfileOut( __METHOD__."-title" );
1611                                 continue;
1612                         }
1613
1614                         $ns = $nt->getNamespace();
1615                         $iw = $nt->getInterWiki();
1616                         wfProfileOut( __METHOD__."-title" );
1617
1618                         if ($might_be_img) { # if this is actually an invalid link
1619                                 wfProfileIn( __METHOD__."-might_be_img" );
1620                                 if ($ns == NS_FILE && $noforce) { #but might be an image
1621                                         $found = false;
1622                                         while ( true ) {
1623                                                 #look at the next 'line' to see if we can close it there
1624                                                 $a->next();
1625                                                 $next_line = $a->current();
1626                                                 if ( $next_line === false || $next_line === null ) {
1627                                                         break;
1628                                                 }
1629                                                 $m = explode( ']]', $next_line, 3 );
1630                                                 if ( count( $m ) == 3 ) {
1631                                                         # the first ]] closes the inner link, the second the image
1632                                                         $found = true;
1633                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1634                                                         $trail = $m[2];
1635                                                         break;
1636                                                 } elseif ( count( $m ) == 2 ) {
1637                                                         #if there's exactly one ]] that's fine, we'll keep looking
1638                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1639                                                 } else {
1640                                                         #if $next_line is invalid too, we need look no further
1641                                                         $text .= '[[' . $next_line;
1642                                                         break;
1643                                                 }
1644                                         }
1645                                         if ( !$found ) {
1646                                                 # we couldn't find the end of this imageLink, so output it raw
1647                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1648                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1649                                                 $s .= "{$prefix}[[$link|$text";
1650                                                 # note: no $trail, because without an end, there *is* no trail
1651                                                 wfProfileOut( __METHOD__."-might_be_img" );
1652                                                 continue;
1653                                         }
1654                                 } else { #it's not an image, so output it raw
1655                                         $s .= "{$prefix}[[$link|$text";
1656                                         # note: no $trail, because without an end, there *is* no trail
1657                                         wfProfileOut( __METHOD__."-might_be_img" );
1658                                         continue;
1659                                 }
1660                                 wfProfileOut( __METHOD__."-might_be_img" );
1661                         }
1662
1663                         $wasblank = ( '' == $text );
1664                         if( $wasblank ) $text = $link;
1665
1666                         # Link not escaped by : , create the various objects
1667                         if( $noforce ) {
1668
1669                                 # Interwikis
1670                                 wfProfileIn( __METHOD__."-interwiki" );
1671                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1672                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1673                                         $s = rtrim($s . $prefix);
1674                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1675                                         wfProfileOut( __METHOD__."-interwiki" );
1676                                         continue;
1677                                 }
1678                                 wfProfileOut( __METHOD__."-interwiki" );
1679
1680                                 if ( $ns == NS_FILE ) {
1681                                         wfProfileIn( __METHOD__."-image" );
1682                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1683                                                 # recursively parse links inside the image caption
1684                                                 # actually, this will parse them in any other parameters, too,
1685                                                 # but it might be hard to fix that, and it doesn't matter ATM
1686                                                 $text = $this->replaceExternalLinks($text);
1687                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1688
1689                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1690                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1691                                         }
1692                                         $this->mOutput->addImage( $nt->getDBkey() );
1693                                         wfProfileOut( __METHOD__."-image" );
1694                                         continue;
1695
1696                                 }
1697
1698                                 if ( $ns == NS_CATEGORY ) {
1699                                         wfProfileIn( __METHOD__."-category" );
1700                                         $s = rtrim($s . "\n"); # bug 87
1701
1702                                         if ( $wasblank ) {
1703                                                 $sortkey = $this->getDefaultSort();
1704                                         } else {
1705                                                 $sortkey = $text;
1706                                         }
1707                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1708                                         $sortkey = str_replace( "\n", '', $sortkey );
1709                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1710                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1711
1712                                         /**
1713                                          * Strip the whitespace Category links produce, see bug 87
1714                                          * @todo We might want to use trim($tmp, "\n") here.
1715                                          */
1716                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1717
1718                                         wfProfileOut( __METHOD__."-category" );
1719                                         continue;
1720                                 }
1721                         }
1722
1723                         # Self-link checking
1724                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1725                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1726                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1727                                         continue;
1728                                 }
1729                         }
1730
1731                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1732                         # FIXME: Should do batch file existence checks, see comment below
1733                         if( $ns == NS_MEDIA ) {
1734                                 wfProfileIn( __METHOD__."-media" );
1735                                 # Give extensions a chance to select the file revision for us
1736                                 $skip = $time = false;
1737                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1738                                 if ( $skip ) {
1739                                         $link = $sk->link( $nt );
1740                                 } else {
1741                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1742                                 }
1743                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1744                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1745                                 $this->mOutput->addImage( $nt->getDBkey() );
1746                                 wfProfileOut( __METHOD__."-media" );
1747                                 continue;
1748                         }
1749
1750                         wfProfileIn( __METHOD__."-always_known" );
1751                         # Some titles, such as valid special pages or files in foreign repos, should
1752                         # be shown as bluelinks even though they're not included in the page table
1753                         #
1754                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1755                         # batch file existence checks for NS_FILE and NS_MEDIA
1756                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1757                                 $this->mOutput->addLink( $nt );
1758                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1759                         } else {
1760                                 # Links will be added to the output link list after checking
1761                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1762                         }
1763                         wfProfileOut( __METHOD__."-always_known" );
1764                 }
1765                 wfProfileOut( __METHOD__ );
1766                 return $holders;
1767         }
1768
1769         /**
1770          * Make a link placeholder. The text returned can be later resolved to a real link with
1771          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1772          * parsing of interwiki links, and secondly to allow all existence checks and
1773          * article length checks (for stub links) to be bundled into a single query.
1774          *
1775          * @deprecated
1776          */
1777         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1778                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1779         }
1780
1781         /**
1782          * Render a forced-blue link inline; protect against double expansion of
1783          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1784          * Since this little disaster has to split off the trail text to avoid
1785          * breaking URLs in the following text without breaking trails on the
1786          * wiki links, it's been made into a horrible function.
1787          *
1788          * @param Title $nt
1789          * @param string $text
1790          * @param string $query
1791          * @param string $trail
1792          * @param string $prefix
1793          * @return string HTML-wikitext mix oh yuck
1794          */
1795         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1796                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1797                 $sk = $this->mOptions->getSkin();
1798                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1799                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1800                 return $this->armorLinks( $link ) . $trail;
1801         }
1802
1803         /**
1804          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1805          * going to go through further parsing steps before inline URL expansion.
1806          *
1807          * Not needed quite as much as it used to be since free links are a bit
1808          * more sensible these days. But bracketed links are still an issue.
1809          *
1810          * @param string more-or-less HTML
1811          * @return string less-or-more HTML with NOPARSE bits
1812          */
1813         function armorLinks( $text ) {
1814                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1815                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1816         }
1817
1818         /**
1819          * Return true if subpage links should be expanded on this page.
1820          * @return bool
1821          */
1822         function areSubpagesAllowed() {
1823                 # Some namespaces don't allow subpages
1824                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1825         }
1826
1827         /**
1828          * Handle link to subpage if necessary
1829          * @param string $target the source of the link
1830          * @param string &$text the link text, modified as necessary
1831          * @return string the full name of the link
1832          * @private
1833          */
1834         function maybeDoSubpageLink($target, &$text) {
1835                 # Valid link forms:
1836                 # Foobar -- normal
1837                 # :Foobar -- override special treatment of prefix (images, language links)
1838                 # /Foobar -- convert to CurrentPage/Foobar
1839                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1840                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1841                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1842
1843                 wfProfileIn( __METHOD__ );
1844                 $ret = $target; # default return value is no change
1845
1846                 # Some namespaces don't allow subpages,
1847                 # so only perform processing if subpages are allowed
1848                 if( $this->areSubpagesAllowed() ) {
1849                         $hash = strpos( $target, '#' );
1850                         if( $hash !== false ) {
1851                                 $suffix = substr( $target, $hash );
1852                                 $target = substr( $target, 0, $hash );
1853                         } else {
1854                                 $suffix = '';
1855                         }
1856                         # bug 7425
1857                         $target = trim( $target );
1858                         # Look at the first character
1859                         if( $target != '' && $target{0} === '/' ) {
1860                                 # / at end means we don't want the slash to be shown
1861                                 $m = array();
1862                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1863                                 if( $trailingSlashes ) {
1864                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1865                                 } else {
1866                                         $noslash = substr( $target, 1 );
1867                                 }
1868
1869                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1870                                 if( '' === $text ) {
1871                                         $text = $target . $suffix;
1872                                 } # this might be changed for ugliness reasons
1873                         } else {
1874                                 # check for .. subpage backlinks
1875                                 $dotdotcount = 0;
1876                                 $nodotdot = $target;
1877                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1878                                         ++$dotdotcount;
1879                                         $nodotdot = substr( $nodotdot, 3 );
1880                                 }
1881                                 if($dotdotcount > 0) {
1882                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1883                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1884                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1885                                                 # / at the end means don't show full path
1886                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1887                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1888                                                         if( '' === $text ) {
1889                                                                 $text = $nodotdot . $suffix;
1890                                                         }
1891                                                 }
1892                                                 $nodotdot = trim( $nodotdot );
1893                                                 if( $nodotdot != '' ) {
1894                                                         $ret .= '/' . $nodotdot;
1895                                                 }
1896                                                 $ret .= $suffix;
1897                                         }
1898                                 }
1899                         }
1900                 }
1901
1902                 wfProfileOut( __METHOD__ );
1903                 return $ret;
1904         }
1905
1906         /**#@+
1907          * Used by doBlockLevels()
1908          * @private
1909          */
1910         /* private */ function closeParagraph() {
1911                 $result = '';
1912                 if ( '' != $this->mLastSection ) {
1913                         $result = '</' . $this->mLastSection  . ">\n";
1914                 }
1915                 $this->mInPre = false;
1916                 $this->mLastSection = '';
1917                 return $result;
1918         }
1919         # getCommon() returns the length of the longest common substring
1920         # of both arguments, starting at the beginning of both.
1921         #
1922         /* private */ function getCommon( $st1, $st2 ) {
1923                 $fl = strlen( $st1 );
1924                 $shorter = strlen( $st2 );
1925                 if ( $fl < $shorter ) { $shorter = $fl; }
1926
1927                 for ( $i = 0; $i < $shorter; ++$i ) {
1928                         if ( $st1{$i} != $st2{$i} ) { break; }
1929                 }
1930                 return $i;
1931         }
1932         # These next three functions open, continue, and close the list
1933         # element appropriate to the prefix character passed into them.
1934         #
1935         /* private */ function openList( $char ) {
1936                 $result = $this->closeParagraph();
1937
1938                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1939                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1940                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1941                 elseif ( ';' === $char ) {
1942                         $result .= '<dl><dt>';
1943                         $this->mDTopen = true;
1944                 }
1945                 else { $result = '<!-- ERR 1 -->'; }
1946
1947                 return $result;
1948         }
1949
1950         /* private */ function nextItem( $char ) {
1951                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1952                 elseif ( ':' === $char || ';' === $char ) {
1953                         $close = '</dd>';
1954                         if ( $this->mDTopen ) { $close = '</dt>'; }
1955                         if ( ';' === $char ) {
1956                                 $this->mDTopen = true;
1957                                 return $close . '<dt>';
1958                         } else {
1959                                 $this->mDTopen = false;
1960                                 return $close . '<dd>';
1961                         }
1962                 }
1963                 return '<!-- ERR 2 -->';
1964         }
1965
1966         /* private */ function closeList( $char ) {
1967                 if ( '*' === $char ) { $text = '</li></ul>'; }
1968                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1969                 elseif ( ':' === $char ) {
1970                         if ( $this->mDTopen ) {
1971                                 $this->mDTopen = false;
1972                                 $text = '</dt></dl>';
1973                         } else {
1974                                 $text = '</dd></dl>';
1975                         }
1976                 }
1977                 else {  return '<!-- ERR 3 -->'; }
1978                 return $text."\n";
1979         }
1980         /**#@-*/
1981
1982         /**
1983          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1984          *
1985          * @param $linestart bool whether or not this is at the start of a line.
1986          * @private
1987          * @return string the lists rendered as HTML
1988          */
1989         function doBlockLevels( $text, $linestart ) {
1990                 wfProfileIn( __METHOD__ );
1991
1992                 # Parsing through the text line by line.  The main thing
1993                 # happening here is handling of block-level elements p, pre,
1994                 # and making lists from lines starting with * # : etc.
1995                 #
1996                 $textLines = StringUtils::explode( "\n", $text );
1997
1998                 $lastPrefix = $output = '';
1999                 $this->mDTopen = $inBlockElem = false;
2000                 $prefixLength = 0;
2001                 $paragraphStack = false;
2002
2003                 foreach ( $textLines as $oLine ) {
2004                         # Fix up $linestart
2005                         if ( !$linestart ) {
2006                                 $output .= $oLine;
2007                                 $linestart = true;
2008                                 continue;
2009                         }
2010                         // * = ul
2011                         // # = ol
2012                         // ; = dt
2013                         // : = dd
2014
2015                         $lastPrefixLength = strlen( $lastPrefix );
2016                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2017                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2018                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2019                         if ( !$this->mInPre ) {
2020                                 # Multiple prefixes may abut each other for nested lists.
2021                                 $prefixLength = strspn( $oLine, '*#:;' );
2022                                 $prefix = substr( $oLine, 0, $prefixLength );
2023
2024                                 # eh?
2025                                 // ; and : are both from definition-lists, so they're equivalent
2026                                 //  for the purposes of determining whether or not we need to open/close
2027                                 //  elements.
2028                                 $prefix2 = str_replace( ';', ':', $prefix );
2029                                 $t = substr( $oLine, $prefixLength );
2030                                 $this->mInPre = (bool)$preOpenMatch;
2031                         } else {
2032                                 # Don't interpret any other prefixes in preformatted text
2033                                 $prefixLength = 0;
2034                                 $prefix = $prefix2 = '';
2035                                 $t = $oLine;
2036                         }
2037
2038                         # List generation
2039                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2040                                 # Same as the last item, so no need to deal with nesting or opening stuff
2041                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2042                                 $paragraphStack = false;
2043
2044                                 if ( substr( $prefix, -1 ) === ';') {
2045                                         # The one nasty exception: definition lists work like this:
2046                                         # ; title : definition text
2047                                         # So we check for : in the remainder text to split up the
2048                                         # title and definition, without b0rking links.
2049                                         $term = $t2 = '';
2050                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2051                                                 $t = $t2;
2052                                                 $output .= $term . $this->nextItem( ':' );
2053                                         }
2054                                 }
2055                         } elseif( $prefixLength || $lastPrefixLength ) {
2056                                 // We need to open or close prefixes, or both.
2057
2058                                 # Either open or close a level...
2059                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2060                                 $paragraphStack = false;
2061
2062                                 // Close all the prefixes which aren't shared.
2063                                 while( $commonPrefixLength < $lastPrefixLength ) {
2064                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2065                                         --$lastPrefixLength;
2066                                 }
2067
2068                                 // Continue the current prefix if appropriate.
2069                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2070                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2071                                 }
2072
2073                                 // Open prefixes where appropriate.
2074                                 while ( $prefixLength > $commonPrefixLength ) {
2075                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2076                                         $output .= $this->openList( $char );
2077
2078                                         if ( ';' === $char ) {
2079                                                 # FIXME: This is dupe of code above
2080                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2081                                                         $t = $t2;
2082                                                         $output .= $term . $this->nextItem( ':' );
2083                                                 }
2084                                         }
2085                                         ++$commonPrefixLength;
2086                                 }
2087                                 $lastPrefix = $prefix2;
2088                         }
2089
2090                         // If we have no prefixes, go to paragraph mode.
2091                         if( 0 == $prefixLength ) {
2092                                 wfProfileIn( __METHOD__."-paragraph" );
2093                                 # No prefix (not in list)--go to paragraph mode
2094                                 // XXX: use a stack for nestable elements like span, table and div
2095                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2096                                 $closematch = preg_match(
2097                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2098                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2099                                 if ( $openmatch or $closematch ) {
2100                                         $paragraphStack = false;
2101                                         # TODO bug 5718: paragraph closed
2102                                         $output .= $this->closeParagraph();
2103                                         if ( $preOpenMatch and !$preCloseMatch ) {
2104                                                 $this->mInPre = true;
2105                                         }
2106                                         if ( $closematch ) {
2107                                                 $inBlockElem = false;
2108                                         } else {
2109                                                 $inBlockElem = true;
2110                                         }
2111                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2112                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2113                                                 // pre
2114                                                 if ($this->mLastSection !== 'pre') {
2115                                                         $paragraphStack = false;
2116                                                         $output .= $this->closeParagraph().'<pre>';
2117                                                         $this->mLastSection = 'pre';
2118                                                 }
2119                                                 $t = substr( $t, 1 );
2120                                         } else {
2121                                                 // paragraph
2122                                                 if ( '' == trim($t) ) {
2123                                                         if ( $paragraphStack ) {
2124                                                                 $output .= $paragraphStack.'<br />';
2125                                                                 $paragraphStack = false;
2126                                                                 $this->mLastSection = 'p';
2127                                                         } else {
2128                                                                 if ($this->mLastSection !== 'p' ) {
2129                                                                         $output .= $this->closeParagraph();
2130                                                                         $this->mLastSection = '';
2131                                                                         $paragraphStack = '<p>';
2132                                                                 } else {
2133                                                                         $paragraphStack = '</p><p>';
2134                                                                 }
2135                                                         }
2136                                                 } else {
2137                                                         if ( $paragraphStack ) {
2138                                                                 $output .= $paragraphStack;
2139                                                                 $paragraphStack = false;
2140                                                                 $this->mLastSection = 'p';
2141                                                         } else if ($this->mLastSection !== 'p') {
2142                                                                 $output .= $this->closeParagraph().'<p>';
2143                                                                 $this->mLastSection = 'p';
2144                                                         }
2145                                                 }
2146                                         }
2147                                 }
2148                                 wfProfileOut( __METHOD__."-paragraph" );
2149                         }
2150                         // somewhere above we forget to get out of pre block (bug 785)
2151                         if($preCloseMatch && $this->mInPre) {
2152                                 $this->mInPre = false;
2153                         }
2154                         if ($paragraphStack === false) {
2155                                 $output .= $t."\n";
2156                         }
2157                 }
2158                 while ( $prefixLength ) {
2159                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2160                         --$prefixLength;
2161                 }
2162                 if ( '' != $this->mLastSection ) {
2163                         $output .= '</' . $this->mLastSection . '>';
2164                         $this->mLastSection = '';
2165                 }
2166
2167                 wfProfileOut( __METHOD__ );
2168                 return $output;
2169         }
2170
2171         /**
2172          * Split up a string on ':', ignoring any occurences inside tags
2173          * to prevent illegal overlapping.
2174          * @param string $str the string to split
2175          * @param string &$before set to everything before the ':'
2176          * @param string &$after set to everything after the ':'
2177          * return string the position of the ':', or false if none found
2178          */
2179         function findColonNoLinks($str, &$before, &$after) {
2180                 wfProfileIn( __METHOD__ );
2181
2182                 $pos = strpos( $str, ':' );
2183                 if( $pos === false ) {
2184                         // Nothing to find!
2185                         wfProfileOut( __METHOD__ );
2186                         return false;
2187                 }
2188
2189                 $lt = strpos( $str, '<' );
2190                 if( $lt === false || $lt > $pos ) {
2191                         // Easy; no tag nesting to worry about
2192                         $before = substr( $str, 0, $pos );
2193                         $after = substr( $str, $pos+1 );
2194                         wfProfileOut( __METHOD__ );
2195                         return $pos;
2196                 }
2197
2198                 // Ugly state machine to walk through avoiding tags.
2199                 $state = self::COLON_STATE_TEXT;
2200                 $stack = 0;
2201                 $len = strlen( $str );
2202                 for( $i = 0; $i < $len; $i++ ) {
2203                         $c = $str{$i};
2204
2205                         switch( $state ) {
2206                         // (Using the number is a performance hack for common cases)
2207                         case 0: // self::COLON_STATE_TEXT:
2208                                 switch( $c ) {
2209                                 case "<":
2210                                         // Could be either a <start> tag or an </end> tag
2211                                         $state = self::COLON_STATE_TAGSTART;
2212                                         break;
2213                                 case ":":
2214                                         if( $stack == 0 ) {
2215                                                 // We found it!
2216                                                 $before = substr( $str, 0, $i );
2217                                                 $after = substr( $str, $i + 1 );
2218                                                 wfProfileOut( __METHOD__ );
2219                                                 return $i;
2220                                         }
2221                                         // Embedded in a tag; don't break it.
2222                                         break;
2223                                 default:
2224                                         // Skip ahead looking for something interesting
2225                                         $colon = strpos( $str, ':', $i );
2226                                         if( $colon === false ) {
2227                                                 // Nothing else interesting
2228                                                 wfProfileOut( __METHOD__ );
2229                                                 return false;
2230                                         }
2231                                         $lt = strpos( $str, '<', $i );
2232                                         if( $stack === 0 ) {
2233                                                 if( $lt === false || $colon < $lt ) {
2234                                                         // We found it!
2235                                                         $before = substr( $str, 0, $colon );
2236                                                         $after = substr( $str, $colon + 1 );
2237                                                         wfProfileOut( __METHOD__ );
2238                                                         return $i;
2239                                                 }
2240                                         }
2241                                         if( $lt === false ) {
2242                                                 // Nothing else interesting to find; abort!
2243                                                 // We're nested, but there's no close tags left. Abort!
2244                                                 break 2;
2245                                         }
2246                                         // Skip ahead to next tag start
2247                                         $i = $lt;
2248                                         $state = self::COLON_STATE_TAGSTART;
2249                                 }
2250                                 break;
2251                         case 1: // self::COLON_STATE_TAG:
2252                                 // In a <tag>
2253                                 switch( $c ) {
2254                                 case ">":
2255                                         $stack++;
2256                                         $state = self::COLON_STATE_TEXT;
2257                                         break;
2258                                 case "/":
2259                                         // Slash may be followed by >?
2260                                         $state = self::COLON_STATE_TAGSLASH;
2261                                         break;
2262                                 default:
2263                                         // ignore
2264                                 }
2265                                 break;
2266                         case 2: // self::COLON_STATE_TAGSTART:
2267                                 switch( $c ) {
2268                                 case "/":
2269                                         $state = self::COLON_STATE_CLOSETAG;
2270                                         break;
2271                                 case "!":
2272                                         $state = self::COLON_STATE_COMMENT;
2273                                         break;
2274                                 case ">":
2275                                         // Illegal early close? This shouldn't happen D:
2276                                         $state = self::COLON_STATE_TEXT;
2277                                         break;
2278                                 default:
2279                                         $state = self::COLON_STATE_TAG;
2280                                 }
2281                                 break;
2282                         case 3: // self::COLON_STATE_CLOSETAG:
2283                                 // In a </tag>
2284                                 if( $c === ">" ) {
2285                                         $stack--;
2286                                         if( $stack < 0 ) {
2287                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2288                                                 wfProfileOut( __METHOD__ );
2289                                                 return false;
2290                                         }
2291                                         $state = self::COLON_STATE_TEXT;
2292                                 }
2293                                 break;
2294                         case self::COLON_STATE_TAGSLASH:
2295                                 if( $c === ">" ) {
2296                                         // Yes, a self-closed tag <blah/>
2297                                         $state = self::COLON_STATE_TEXT;
2298                                 } else {
2299                                         // Probably we're jumping the gun, and this is an attribute
2300                                         $state = self::COLON_STATE_TAG;
2301                                 }
2302                                 break;
2303                         case 5: // self::COLON_STATE_COMMENT:
2304                                 if( $c === "-" ) {
2305                                         $state = self::COLON_STATE_COMMENTDASH;
2306                                 }
2307                                 break;
2308                         case self::COLON_STATE_COMMENTDASH:
2309                                 if( $c === "-" ) {
2310                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2311                                 } else {
2312                                         $state = self::COLON_STATE_COMMENT;
2313                                 }
2314                                 break;
2315                         case self::COLON_STATE_COMMENTDASHDASH:
2316                                 if( $c === ">" ) {
2317                                         $state = self::COLON_STATE_TEXT;
2318                                 } else {
2319                                         $state = self::COLON_STATE_COMMENT;
2320                                 }
2321                                 break;
2322                         default:
2323                                 throw new MWException( "State machine error in " . __METHOD__ );
2324                         }
2325                 }
2326                 if( $stack > 0 ) {
2327                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2328                         return false;
2329                 }
2330                 wfProfileOut( __METHOD__ );
2331                 return false;
2332         }
2333
2334         /**
2335          * Return value of a magic variable (like PAGENAME)
2336          *
2337          * @private
2338          */
2339         function getVariableValue( $index ) {
2340                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2341
2342                 /**
2343                  * Some of these require message or data lookups and can be
2344                  * expensive to check many times.
2345                  */
2346                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2347                         if ( isset( $this->mVarCache[$index] ) ) {
2348                                 return $this->mVarCache[$index];
2349                         }
2350                 }
2351
2352                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2353                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2354
2355                 # Use the time zone
2356                 global $wgLocaltimezone;
2357                 if ( isset( $wgLocaltimezone ) ) {
2358                         $oldtz = getenv( 'TZ' );
2359                         putenv( 'TZ='.$wgLocaltimezone );
2360                 }
2361
2362                 wfSuppressWarnings(); // E_STRICT system time bitching
2363                 $localTimestamp = date( 'YmdHis', $ts );
2364                 $localMonth = date( 'm', $ts );
2365                 $localMonth1 = date( 'n', $ts );
2366                 $localMonthName = date( 'n', $ts );
2367                 $localDay = date( 'j', $ts );
2368                 $localDay2 = date( 'd', $ts );
2369                 $localDayOfWeek = date( 'w', $ts );
2370                 $localWeek = date( 'W', $ts );
2371                 $localYear = date( 'Y', $ts );
2372                 $localHour = date( 'H', $ts );
2373                 if ( isset( $wgLocaltimezone ) ) {
2374                         putenv( 'TZ='.$oldtz );
2375                 }
2376                 wfRestoreWarnings();
2377
2378                 switch ( $index ) {
2379                         case 'currentmonth':
2380                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2381                         case 'currentmonth1':
2382                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2383                         case 'currentmonthname':
2384                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2385                         case 'currentmonthnamegen':
2386                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2387                         case 'currentmonthabbrev':
2388                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2389                         case 'currentday':
2390                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2391                         case 'currentday2':
2392                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2393                         case 'localmonth':
2394                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2395                         case 'localmonth1':
2396                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth1 );
2397                         case 'localmonthname':
2398                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2399                         case 'localmonthnamegen':
2400                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2401                         case 'localmonthabbrev':
2402                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2403                         case 'localday':
2404                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2405                         case 'localday2':
2406                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2407                         case 'pagename':
2408                                 return wfEscapeWikiText( $this->mTitle->getText() );
2409                         case 'pagenamee':
2410                                 return $this->mTitle->getPartialURL();
2411                         case 'fullpagename':
2412                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2413                         case 'fullpagenamee':
2414                                 return $this->mTitle->getPrefixedURL();
2415                         case 'subpagename':
2416                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2417                         case 'subpagenamee':
2418                                 return $this->mTitle->getSubpageUrlForm();
2419                         case 'basepagename':
2420                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2421                         case 'basepagenamee':
2422                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2423                         case 'talkpagename':
2424                                 if( $this->mTitle->canTalk() ) {
2425                                         $talkPage = $this->mTitle->getTalkPage();
2426                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2427                                 } else {
2428                                         return '';
2429                                 }
2430                         case 'talkpagenamee':
2431                                 if( $this->mTitle->canTalk() ) {
2432                                         $talkPage = $this->mTitle->getTalkPage();
2433                                         return $talkPage->getPrefixedUrl();
2434                                 } else {
2435                                         return '';
2436                                 }
2437                         case 'subjectpagename':
2438                                 $subjPage = $this->mTitle->getSubjectPage();
2439                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2440                         case 'subjectpagenamee':
2441                                 $subjPage = $this->mTitle->getSubjectPage();
2442                                 return $subjPage->getPrefixedUrl();
2443                         case 'revisionid':
2444                                 // Let the edit saving system know we should parse the page
2445                                 // *after* a revision ID has been assigned.
2446                                 $this->mOutput->setFlag( 'vary-revision' );
2447                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2448                                 return $this->mRevisionId;
2449                         case 'revisionday':
2450                                 // Let the edit saving system know we should parse the page
2451                                 // *after* a revision ID has been assigned. This is for null edits.
2452                                 $this->mOutput->setFlag( 'vary-revision' );
2453                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2454                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2455                         case 'revisionday2':
2456                                 // Let the edit saving system know we should parse the page
2457                                 // *after* a revision ID has been assigned. This is for null edits.
2458                                 $this->mOutput->setFlag( 'vary-revision' );
2459                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2460                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2461                         case 'revisionmonth':
2462                                 // Let the edit saving system know we should parse the page
2463                                 // *after* a revision ID has been assigned. This is for null edits.
2464                                 $this->mOutput->setFlag( 'vary-revision' );
2465                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2466                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2467                         case 'revisionyear':
2468                                 // Let the edit saving system know we should parse the page
2469                                 // *after* a revision ID has been assigned. This is for null edits.
2470                                 $this->mOutput->setFlag( 'vary-revision' );
2471                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2472                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2473                         case 'revisiontimestamp':
2474                                 // Let the edit saving system know we should parse the page
2475                                 // *after* a revision ID has been assigned. This is for null edits.
2476                                 $this->mOutput->setFlag( 'vary-revision' );
2477                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2478                                 return $this->getRevisionTimestamp();
2479                         case 'revisionuser':
2480                                 // Let the edit saving system know we should parse the page
2481                                 // *after* a revision ID has been assigned. This is for null edits.
2482                                 $this->mOutput->setFlag( 'vary-revision' );
2483                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2484                                 return $this->getRevisionUser();
2485                         case 'namespace':
2486                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2487                         case 'namespacee':
2488                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2489                         case 'talkspace':
2490                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2491                         case 'talkspacee':
2492                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2493                         case 'subjectspace':
2494                                 return $this->mTitle->getSubjectNsText();
2495                         case 'subjectspacee':
2496                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2497                         case 'currentdayname':
2498                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2499                         case 'currentyear':
2500                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2501                         case 'currenttime':
2502                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2503                         case 'currenthour':
2504                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2505                         case 'currentweek':
2506                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2507                                 // int to remove the padding
2508                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2509                         case 'currentdow':
2510                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2511                         case 'localdayname':
2512                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2513                         case 'localyear':
2514                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2515                         case 'localtime':
2516                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2517                         case 'localhour':
2518                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2519                         case 'localweek':
2520                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2521                                 // int to remove the padding
2522                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2523                         case 'localdow':
2524                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2525                         case 'numberofarticles':
2526                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2527                         case 'numberoffiles':
2528                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2529                         case 'numberofusers':
2530                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2531                         case 'numberofactiveusers':
2532                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::activeUsers() );
2533                         case 'numberofpages':
2534                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2535                         case 'numberofadmins':
2536                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2537                         case 'numberofedits':
2538                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2539                         case 'numberofviews':
2540                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2541                         case 'currenttimestamp':
2542                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2543                         case 'localtimestamp':
2544                                 return $this->mVarCache[$index] = $localTimestamp;
2545                         case 'currentversion':
2546                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2547                         case 'sitename':
2548                                 return $wgSitename;
2549                         case 'server':
2550                                 return $wgServer;
2551                         case 'servername':
2552                                 return $wgServerName;
2553                         case 'scriptpath':
2554                                 return $wgScriptPath;
2555                         case 'directionmark':
2556                                 return $wgContLang->getDirMark();
2557                         case 'contentlanguage':
2558                                 global $wgContLanguageCode;
2559                                 return $wgContLanguageCode;
2560                         default:
2561                                 $ret = null;
2562                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2563                                         return $ret;
2564                                 else
2565                                         return null;
2566                 }
2567         }
2568
2569         /**
2570          * initialise the magic variables (like CURRENTMONTHNAME)
2571          *
2572          * @private
2573          */
2574         function initialiseVariables() {
2575                 wfProfileIn( __METHOD__ );
2576                 $variableIDs = MagicWord::getVariableIDs();
2577
2578                 $this->mVariables = new MagicWordArray( $variableIDs );
2579                 wfProfileOut( __METHOD__ );
2580         }
2581
2582         /**
2583          * Preprocess some wikitext and return the document tree.
2584          * This is the ghost of replace_variables().
2585          *
2586          * @param string $text The text to parse
2587          * @param integer flags Bitwise combination of:
2588          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2589          *                                     included. Default is to assume a direct page view.
2590          *
2591          * The generated DOM tree must depend only on the input text and the flags.
2592          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2593          *
2594          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2595          * change in the DOM tree for a given text, must be passed through the section identifier
2596          * in the section edit link and thus back to extractSections().
2597          *
2598          * The output of this function is currently only cached in process memory, but a persistent
2599          * cache may be implemented at a later date which takes further advantage of these strict
2600          * dependency requirements.
2601          *
2602          * @private
2603          */
2604         function preprocessToDom ( $text, $flags = 0 ) {
2605                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2606                 return $dom;
2607         }
2608
2609         /*
2610          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2611          */
2612         public static function splitWhitespace( $s ) {
2613                 $ltrimmed = ltrim( $s );
2614                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2615                 $trimmed = rtrim( $ltrimmed );
2616                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2617                 if ( $diff > 0 ) {
2618                         $w2 = substr( $ltrimmed, -$diff );
2619                 } else {
2620                         $w2 = '';
2621                 }
2622                 return array( $w1, $trimmed, $w2 );
2623         }
2624
2625         /**
2626          * Replace magic variables, templates, and template arguments
2627          * with the appropriate text. Templates are substituted recursively,
2628          * taking care to avoid infinite loops.
2629          *
2630          * Note that the substitution depends on value of $mOutputType:
2631          *  self::OT_WIKI: only {{subst:}} templates
2632          *  self::OT_PREPROCESS: templates but not extension tags
2633          *  self::OT_HTML: all templates and extension tags
2634          *
2635          * @param string $tex The text to transform
2636          * @param PPFrame $frame Object describing the arguments passed to the template.
2637          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2638          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2639          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2640          * @private
2641          */
2642         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2643                 # Is there any text? Also, Prevent too big inclusions!
2644                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2645                         return $text;
2646                 }
2647                 wfProfileIn( __METHOD__ );
2648
2649                 if ( $frame === false ) {
2650                         $frame = $this->getPreprocessor()->newFrame();
2651                 } elseif ( !( $frame instanceof PPFrame ) ) {
2652                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2653                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2654                 }
2655
2656                 $dom = $this->preprocessToDom( $text );
2657                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2658                 $text = $frame->expand( $dom, $flags );
2659
2660                 wfProfileOut( __METHOD__ );
2661                 return $text;
2662         }
2663
2664         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2665         static function createAssocArgs( $args ) {
2666                 $assocArgs = array();
2667                 $index = 1;
2668                 foreach( $args as $arg ) {
2669                         $eqpos = strpos( $arg, '=' );
2670                         if ( $eqpos === false ) {
2671                                 $assocArgs[$index++] = $arg;
2672                         } else {
2673                                 $name = trim( substr( $arg, 0, $eqpos ) );
2674                                 $value = trim( substr( $arg, $eqpos+1 ) );
2675                                 if ( $value === false ) {
2676                                         $value = '';
2677                                 }
2678                                 if ( $name !== false ) {
2679                                         $assocArgs[$name] = $value;
2680                                 }
2681                         }
2682                 }
2683
2684                 return $assocArgs;
2685         }
2686
2687         /**
2688          * Warn the user when a parser limitation is reached
2689          * Will warn at most once the user per limitation type
2690          *
2691          * @param string $limitationType, should be one of:
2692          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2693          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2694          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2695          * @params int $current, $max When an explicit limit has been
2696          *       exceeded, provide the values (optional)
2697          */
2698         function limitationWarn( $limitationType, $current=null, $max=null) {
2699                 $msgName = $limitationType . '-warning';
2700                 //does no harm if $current and $max are present but are unnecessary for the message
2701                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2702                 $this->mOutput->addWarning( $warning );
2703                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2704                 if ( $cat ) {
2705                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2706                 }
2707         }
2708
2709         /**
2710          * Return the text of a template, after recursively
2711          * replacing any variables or templates within the template.
2712          *
2713          * @param array $piece The parts of the template
2714          *  $piece['title']: the title, i.e. the part before the |
2715          *  $piece['parts']: the parameter array
2716          *  $piece['lineStart']: whether the brace was at the start of a line
2717          * @param PPFrame The current frame, contains template arguments
2718          * @return string the text of the template
2719          * @private
2720          */
2721         function braceSubstitution( $piece, $frame ) {
2722                 global $wgContLang, $wgNonincludableNamespaces;
2723                 wfProfileIn( __METHOD__ );
2724                 wfProfileIn( __METHOD__.'-setup' );
2725
2726                 # Flags
2727                 $found = false;             # $text has been filled
2728                 $nowiki = false;            # wiki markup in $text should be escaped
2729                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2730                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2731                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2732                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2733
2734                 # Title object, where $text came from
2735                 $title = NULL;
2736
2737                 # $part1 is the bit before the first |, and must contain only title characters.
2738                 # Various prefixes will be stripped from it later.
2739                 $titleWithSpaces = $frame->expand( $piece['title'] );
2740                 $part1 = trim( $titleWithSpaces );
2741                 $titleText = false;
2742
2743                 # Original title text preserved for various purposes
2744                 $originalTitle = $part1;
2745
2746                 # $args is a list of argument nodes, starting from index 0, not including $part1
2747                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2748                 wfProfileOut( __METHOD__.'-setup' );
2749
2750                 # SUBST
2751                 wfProfileIn( __METHOD__.'-modifiers' );
2752                 if ( !$found ) {
2753                         $mwSubst = MagicWord::get( 'subst' );
2754                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2755                                 # One of two possibilities is true:
2756                                 # 1) Found SUBST but not in the PST phase
2757                                 # 2) Didn't find SUBST and in the PST phase
2758                                 # In either case, return without further processing
2759                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2760                                 $isLocalObj = true;
2761                                 $found = true;
2762                         }
2763                 }
2764
2765                 # Variables
2766                 if ( !$found && $args->getLength() == 0 ) {
2767                         $id = $this->mVariables->matchStartToEnd( $part1 );
2768                         if ( $id !== false ) {
2769                                 $text = $this->getVariableValue( $id );
2770                                 if (MagicWord::getCacheTTL($id)>-1)
2771                                         $this->mOutput->mContainsOldMagic = true;
2772                                 $found = true;
2773                         }
2774                 }
2775
2776                 # MSG, MSGNW and RAW
2777                 if ( !$found ) {
2778                         # Check for MSGNW:
2779                         $mwMsgnw = MagicWord::get( 'msgnw' );
2780                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2781                                 $nowiki = true;
2782                         } else {
2783                                 # Remove obsolete MSG:
2784                                 $mwMsg = MagicWord::get( 'msg' );
2785                                 $mwMsg->matchStartAndRemove( $part1 );
2786                         }
2787
2788                         # Check for RAW:
2789                         $mwRaw = MagicWord::get( 'raw' );
2790                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2791                                 $forceRawInterwiki = true;
2792                         }
2793                 }
2794                 wfProfileOut( __METHOD__.'-modifiers' );
2795
2796                 # Parser functions
2797                 if ( !$found ) {
2798                         wfProfileIn( __METHOD__ . '-pfunc' );
2799
2800                         $colonPos = strpos( $part1, ':' );
2801                         if ( $colonPos !== false ) {
2802                                 # Case sensitive functions
2803                                 $function = substr( $part1, 0, $colonPos );
2804                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2805                                         $function = $this->mFunctionSynonyms[1][$function];
2806                                 } else {
2807                                         # Case insensitive functions
2808                                         $function = strtolower( $function );
2809                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2810                                                 $function = $this->mFunctionSynonyms[0][$function];
2811                                         } else {
2812                                                 $function = false;
2813                                         }
2814                                 }
2815                                 if ( $function ) {
2816                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2817                                         $initialArgs = array( &$this );
2818                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2819                                         if ( $flags & SFH_OBJECT_ARGS ) {
2820                                                 # Add a frame parameter, and pass the arguments as an array
2821                                                 $allArgs = $initialArgs;
2822                                                 $allArgs[] = $frame;
2823                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2824                                                         $funcArgs[] = $args->item( $i );
2825                                                 }
2826                                                 $allArgs[] = $funcArgs;
2827                                         } else {
2828                                                 # Convert arguments to plain text
2829                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2830                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2831                                                 }
2832                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2833                                         }
2834
2835                                         # Workaround for PHP bug 35229 and similar
2836                                         if ( !is_callable( $callback ) ) {
2837                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2838                                                 wfProfileOut( __METHOD__ );
2839                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2840                                         }
2841                                         $result = call_user_func_array( $callback, $allArgs );
2842                                         $found = true;
2843                                         $noparse = true;
2844                                         $preprocessFlags = 0;
2845
2846                                         if ( is_array( $result ) ) {
2847                                                 if ( isset( $result[0] ) ) {
2848                                                         $text = $result[0];
2849                                                         unset( $result[0] );
2850                                                 }
2851
2852                                                 // Extract flags into the local scope
2853                                                 // This allows callers to set flags such as nowiki, found, etc.
2854                                                 extract( $result );
2855                                         } else {
2856                                                 $text = $result;
2857                                         }
2858                                         if ( !$noparse ) {
2859                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2860                                                 $isChildObj = true;
2861                                         }
2862                                 }
2863                         }
2864                         wfProfileOut( __METHOD__ . '-pfunc' );
2865                 }
2866
2867                 # Finish mangling title and then check for loops.
2868                 # Set $title to a Title object and $titleText to the PDBK
2869                 if ( !$found ) {
2870                         $ns = NS_TEMPLATE;
2871                         # Split the title into page and subpage
2872                         $subpage = '';
2873                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2874                         if ($subpage !== '') {
2875                                 $ns = $this->mTitle->getNamespace();
2876                         }
2877                         $title = Title::newFromText( $part1, $ns );
2878                         if ( $title ) {
2879                                 $titleText = $title->getPrefixedText();
2880                                 # Check for language variants if the template is not found
2881                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2882                                         $wgContLang->findVariantLink( $part1, $title, true );
2883                                 }
2884                                 # Do recursion depth check
2885                                 $limit = $this->mOptions->getMaxTemplateDepth();
2886                                 if ( $frame->depth >= $limit ) {
2887                                         $found = true;
2888                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2889                                 }
2890                         }
2891                 }
2892
2893                 # Load from database
2894                 if ( !$found && $title ) {
2895                         wfProfileIn( __METHOD__ . '-loadtpl' );
2896                         if ( !$title->isExternal() ) {
2897                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2898                                         $text = SpecialPage::capturePath( $title );
2899                                         if ( is_string( $text ) ) {
2900                                                 $found = true;
2901                                                 $isHTML = true;
2902                                                 $this->disableCache();
2903                                         }
2904                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2905                                         $found = false; //access denied
2906                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2907                                 } else {
2908                                         list( $text, $title ) = $this->getTemplateDom( $title );
2909                                         if ( $text !== false ) {
2910                                                 $found = true;
2911                                                 $isChildObj = true;
2912                                         }
2913                                 }
2914
2915                                 # If the title is valid but undisplayable, make a link to it
2916                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2917                                         $text = "[[:$titleText]]";
2918                                         $found = true;
2919                                 }
2920                         } elseif ( $title->isTrans() ) {
2921                                 // Interwiki transclusion
2922                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2923                                         $text = $this->interwikiTransclude( $title, 'render' );
2924                                         $isHTML = true;
2925                                 } else {
2926                                         $text = $this->interwikiTransclude( $title, 'raw' );
2927                                         // Preprocess it like a template
2928                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2929                                         $isChildObj = true;
2930                                 }
2931                                 $found = true;
2932                         }
2933
2934                         # Do infinite loop check
2935                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2936                         if ( !$frame->loopCheck( $title ) ) {
2937                                 $found = true;
2938                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2939                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2940                         }
2941                         wfProfileOut( __METHOD__ . '-loadtpl' );
2942                 }
2943
2944                 # If we haven't found text to substitute by now, we're done
2945                 # Recover the source wikitext and return it
2946                 if ( !$found ) {
2947                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2948                         wfProfileOut( __METHOD__ );
2949                         return array( 'object' => $text );
2950                 }
2951
2952                 # Expand DOM-style return values in a child frame
2953                 if ( $isChildObj ) {
2954                         # Clean up argument array
2955                         $newFrame = $frame->newChild( $args, $title );
2956
2957                         if ( $nowiki ) {
2958                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2959                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2960                                 # Expansion is eligible for the empty-frame cache
2961                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2962                                         $text = $this->mTplExpandCache[$titleText];
2963                                 } else {
2964                                         $text = $newFrame->expand( $text );
2965                                         $this->mTplExpandCache[$titleText] = $text;
2966                                 }
2967                         } else {
2968                                 # Uncached expansion
2969                                 $text = $newFrame->expand( $text );
2970                         }
2971                 }
2972                 if ( $isLocalObj && $nowiki ) {
2973                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2974                         $isLocalObj = false;
2975                 }
2976
2977                 # Replace raw HTML by a placeholder
2978                 # Add a blank line preceding, to prevent it from mucking up
2979                 # immediately preceding headings
2980                 if ( $isHTML ) {
2981                         $text = "\n\n" . $this->insertStripItem( $text );
2982                 }
2983                 # Escape nowiki-style return values
2984                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2985                         $text = wfEscapeWikiText( $text );
2986                 }
2987                 # Bug 529: if the template begins with a table or block-level
2988                 # element, it should be treated as beginning a new line.
2989                 # This behaviour is somewhat controversial.
2990                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2991                         $text = "\n" . $text;
2992                 }
2993
2994                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2995                         # Error, oversize inclusion
2996                         $text = "[[$originalTitle]]" .
2997                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2998                         $this->limitationWarn( 'post-expand-template-inclusion' );
2999                 }
3000
3001                 if ( $isLocalObj ) {
3002                         $ret = array( 'object' => $text );
3003                 } else {
3004                         $ret = array( 'text' => $text );
3005                 }
3006
3007                 wfProfileOut( __METHOD__ );
3008                 return $ret;
3009         }
3010
3011         /**
3012          * Get the semi-parsed DOM representation of a template with a given title,
3013          * and its redirect destination title. Cached.
3014          */
3015         function getTemplateDom( $title ) {
3016                 $cacheTitle = $title;
3017                 $titleText = $title->getPrefixedDBkey();
3018
3019                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3020                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3021                         $title = Title::makeTitle( $ns, $dbk );
3022                         $titleText = $title->getPrefixedDBkey();
3023                 }
3024                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3025                         return array( $this->mTplDomCache[$titleText], $title );
3026                 }
3027
3028                 // Cache miss, go to the database
3029                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3030
3031                 if ( $text === false ) {
3032                         $this->mTplDomCache[$titleText] = false;
3033                         return array( false, $title );
3034                 }
3035
3036                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3037                 $this->mTplDomCache[ $titleText ] = $dom;
3038
3039                 if (! $title->equals($cacheTitle)) {
3040                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3041                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3042                 }
3043
3044                 return array( $dom, $title );
3045         }
3046
3047         /**
3048          * Fetch the unparsed text of a template and register a reference to it.
3049          */
3050         function fetchTemplateAndTitle( $title ) {
3051                 $templateCb = $this->mOptions->getTemplateCallback();
3052                 $stuff = call_user_func( $templateCb, $title, $this );
3053                 $text = $stuff['text'];
3054                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3055                 if ( isset( $stuff['deps'] ) ) {
3056                         foreach ( $stuff['deps'] as $dep ) {
3057                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3058                         }
3059                 }
3060                 return array($text,$finalTitle);
3061         }
3062
3063         function fetchTemplate( $title ) {
3064                 $rv = $this->fetchTemplateAndTitle($title);
3065                 return $rv[0];
3066         }
3067
3068         /**
3069          * Static function to get a template
3070          * Can be overridden via ParserOptions::setTemplateCallback().
3071          */
3072         static function statelessFetchTemplate( $title, $parser=false ) {
3073                 $text = $skip = false;
3074                 $finalTitle = $title;
3075                 $deps = array();
3076
3077                 // Loop to fetch the article, with up to 1 redirect
3078                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3079                         # Give extensions a chance to select the revision instead
3080                         $id = false; // Assume current
3081                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3082
3083                         if( $skip ) {
3084                                 $text = false;
3085                                 $deps[] = array(
3086                                         'title' => $title,
3087                                         'page_id' => $title->getArticleID(),
3088                                         'rev_id' => null );
3089                                 break;
3090                         }
3091                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3092                         $rev_id = $rev ? $rev->getId() : 0;
3093                         // If there is no current revision, there is no page
3094                         if( $id === false && !$rev ) {
3095                                 $linkCache = LinkCache::singleton();
3096                                 $linkCache->addBadLinkObj( $title );
3097                         }
3098
3099                         $deps[] = array(
3100                                 'title' => $title,
3101                                 'page_id' => $title->getArticleID(),
3102                                 'rev_id' => $rev_id );
3103
3104                         if( $rev ) {
3105                                 $text = $rev->getText();
3106                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3107                                 global $wgContLang;
3108                                 $message = $wgContLang->lcfirst( $title->getText() );
3109                                 $text = wfMsgForContentNoTrans( $message );
3110                                 if( wfEmptyMsg( $message, $text ) ) {
3111                                         $text = false;
3112                                         break;
3113                                 }
3114                         } else {
3115                                 break;
3116                         }
3117                         if ( $text === false ) {
3118                                 break;
3119                         }
3120                         // Redirect?
3121                         $finalTitle = $title;
3122                         $title = Title::newFromRedirect( $text );
3123                 }
3124                 return array(
3125                         'text' => $text,
3126                         'finalTitle' => $finalTitle,
3127                         'deps' => $deps );
3128         }
3129
3130         /**
3131          * Transclude an interwiki link.
3132          */
3133         function interwikiTransclude( $title, $action ) {
3134                 global $wgEnableScaryTranscluding;
3135
3136                 if (!$wgEnableScaryTranscluding)
3137                         return wfMsg('scarytranscludedisabled');
3138
3139                 $url = $title->getFullUrl( "action=$action" );
3140
3141                 if (strlen($url) > 255)
3142                         return wfMsg('scarytranscludetoolong');
3143                 return $this->fetchScaryTemplateMaybeFromCache($url);
3144         }
3145
3146         function fetchScaryTemplateMaybeFromCache($url) {
3147                 global $wgTranscludeCacheExpiry;
3148                 $dbr = wfGetDB(DB_SLAVE);
3149                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3150                                 array('tc_url' => $url));
3151                 if ($obj) {
3152                         $time = $obj->tc_time;
3153                         $text = $obj->tc_contents;
3154                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3155                                 return $text;
3156                         }
3157                 }
3158
3159                 $text = Http::get($url);
3160                 if (!$text)
3161                         return wfMsg('scarytranscludefailed', $url);
3162
3163                 $dbw = wfGetDB(DB_MASTER);
3164                 $dbw->replace('transcache', array('tc_url'), array(
3165                         'tc_url' => $url,
3166                         'tc_time' => time(),
3167                         'tc_contents' => $text));
3168                 return $text;
3169         }
3170
3171
3172         /**
3173          * Triple brace replacement -- used for template arguments
3174          * @private
3175          */
3176         function argSubstitution( $piece, $frame ) {
3177                 wfProfileIn( __METHOD__ );
3178
3179                 $error = false;
3180                 $parts = $piece['parts'];
3181                 $nameWithSpaces = $frame->expand( $piece['title'] );
3182                 $argName = trim( $nameWithSpaces );
3183                 $object = false;
3184                 $text = $frame->getArgument( $argName );
3185                 if (  $text === false && $parts->getLength() > 0
3186                   && (
3187                     $this->ot['html']
3188                     || $this->ot['pre']
3189                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3190                   )
3191                 ) {
3192                         # No match in frame, use the supplied default
3193                         $object = $parts->item( 0 )->getChildren();
3194                 }
3195                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3196                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3197                         $this->limitationWarn( 'post-expand-template-argument' );
3198                 }
3199
3200                 if ( $text === false && $object === false ) {
3201                         # No match anywhere
3202                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3203                 }
3204                 if ( $error !== false ) {
3205                         $text .= $error;
3206                 }
3207                 if ( $object !== false ) {
3208                         $ret = array( 'object' => $object );
3209                 } else {
3210                         $ret = array( 'text' => $text );
3211                 }
3212
3213                 wfProfileOut( __METHOD__ );
3214                 return $ret;
3215         }
3216
3217         /**
3218          * Return the text to be used for a given extension tag.
3219          * This is the ghost of strip().
3220          *
3221          * @param array $params Associative array of parameters:
3222          *     name       PPNode for the tag name
3223          *     attr       PPNode for unparsed text where tag attributes are thought to be
3224          *     attributes Optional associative array of parsed attributes
3225          *     inner      Contents of extension element
3226          *     noClose    Original text did not have a close tag
3227          * @param PPFrame $frame
3228          */
3229         function extensionSubstitution( $params, $frame ) {
3230                 global $wgRawHtml, $wgContLang;
3231
3232                 $name = $frame->expand( $params['name'] );
3233                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3234                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3235
3236                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3237
3238                 if ( $this->ot['html'] ) {
3239                         $name = strtolower( $name );
3240
3241                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3242                         if ( isset( $params['attributes'] ) ) {
3243                                 $attributes = $attributes + $params['attributes'];
3244                         }
3245                         switch ( $name ) {
3246                                 case 'html':
3247                                         if( $wgRawHtml ) {
3248                                                 $output = $content;
3249                                                 break;
3250                                         } else {
3251                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3252                                         }
3253                                 case 'nowiki':
3254                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3255                                         $output = Xml::escapeTagsOnly( $content );
3256                                         break;
3257                                 case 'math':
3258                                         $output = $wgContLang->armourMath(
3259                                                 MathRenderer::renderMath( $content, $attributes ) );
3260                                         break;
3261                                 case 'gallery':
3262                                         $output = $this->renderImageGallery( $content, $attributes );
3263                                         break;
3264                                 default:
3265                                         if( isset( $this->mTagHooks[$name] ) ) {
3266                                                 # Workaround for PHP bug 35229 and similar
3267                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3268                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3269                                                 }
3270                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3271                                                         array( $content, $attributes, $this ) );
3272                                         } else {
3273                                                 $output = '<span class="error">Invalid tag extension name: ' .
3274                                                         htmlspecialchars( $name ) . '</span>';
3275                                         }
3276                         }
3277                 } else {
3278                         if ( is_null( $attrText ) ) {
3279                                 $attrText = '';
3280                         }
3281                         if ( isset( $params['attributes'] ) ) {
3282                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3283                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3284                                                 htmlspecialchars( $attrValue ) . '"';
3285                                 }
3286                         }
3287                         if ( $content === null ) {
3288                                 $output = "<$name$attrText/>";
3289                         } else {
3290                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3291                                 $output = "<$name$attrText>$content$close";
3292                         }
3293                 }
3294
3295                 if ( $name === 'html' || $name === 'nowiki' ) {
3296                         $this->mStripState->nowiki->setPair( $marker, $output );
3297                 } else {
3298                         $this->mStripState->general->setPair( $marker, $output );
3299                 }
3300                 return $marker;
3301         }
3302
3303         /**
3304          * Increment an include size counter
3305          *
3306          * @param string $type The type of expansion
3307          * @param integer $size The size of the text
3308          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3309          */
3310         function incrementIncludeSize( $type, $size ) {
3311                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3312                         return false;
3313                 } else {
3314                         $this->mIncludeSizes[$type] += $size;
3315                         return true;
3316                 }
3317         }
3318
3319         /**
3320          * Increment the expensive function count
3321          *
3322          * @return boolean False if the limit has been exceeded
3323          */
3324         function incrementExpensiveFunctionCount() {
3325                 global $wgExpensiveParserFunctionLimit;
3326                 $this->mExpensiveFunctionCount++;
3327                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3328                         return true;
3329                 }
3330                 return false;
3331         }
3332
3333         /**
3334          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3335          * Fills $this->mDoubleUnderscores, returns the modified text
3336          */
3337         function doDoubleUnderscore( $text ) {
3338                 wfProfileIn( __METHOD__ );
3339                 // The position of __TOC__ needs to be recorded
3340                 $mw = MagicWord::get( 'toc' );
3341                 if( $mw->match( $text ) ) {
3342                         $this->mShowToc = true;
3343                         $this->mForceTocPosition = true;
3344
3345                         // Set a placeholder. At the end we'll fill it in with the TOC.
3346                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3347
3348                         // Only keep the first one.
3349                         $text = $mw->replace( '', $text );
3350                 }
3351
3352                 // Now match and remove the rest of them
3353                 $mwa = MagicWord::getDoubleUnderscoreArray();
3354                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3355
3356                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3357                         $this->mOutput->mNoGallery = true;
3358                 }
3359                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3360                         $this->mShowToc = false;
3361                 }
3362                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3363                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3364
3365                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3366                         if ( $containerCategory ) {
3367                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3368                         } else {
3369                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3370                         }
3371                 }
3372                 # (bug 8068) Allow control over whether robots index a page.
3373                 #
3374                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3375                 # is not desirable, the last one on the page should win.
3376                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3377                         $this->mOutput->setIndexPolicy( 'noindex' );
3378                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3379                         $this->mOutput->setIndexPolicy( 'index' );
3380                 }
3381                 wfProfileOut( __METHOD__ );
3382                 return $text;
3383         }
3384
3385         /**
3386          * This function accomplishes several tasks:
3387          * 1) Auto-number headings if that option is enabled
3388          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3389          * 3) Add a Table of contents on the top for users who have enabled the option
3390          * 4) Auto-anchor headings
3391          *
3392          * It loops through all headlines, collects the necessary data, then splits up the
3393          * string and re-inserts the newly formatted headlines.
3394          *
3395          * @param string $text
3396          * @param string $origText Original, untouched wikitext
3397          * @param boolean $isMain
3398          * @private
3399          */
3400         function formatHeadings( $text, $origText, $isMain=true ) {
3401                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3402
3403                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3404                 $showEditLink = $this->mOptions->getEditSection();
3405
3406                 // Do not call quickUserCan unless necessary
3407                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3408                         $showEditLink = 0;
3409                 }
3410
3411                 # Inhibit editsection links if requested in the page
3412                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3413                         $showEditLink = 0;
3414                 }
3415
3416                 # Get all headlines for numbering them and adding funky stuff like [edit]
3417                 # links - this is for later, but we need the number of headlines right now
3418                 $matches = array();
3419                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3420
3421                 # if there are fewer than 4 headlines in the article, do not show TOC
3422                 # unless it's been explicitly enabled.
3423                 $enoughToc = $this->mShowToc &&
3424                         (($numMatches >= 4) || $this->mForceTocPosition);
3425
3426                 # Allow user to stipulate that a page should have a "new section"
3427                 # link added via __NEWSECTIONLINK__
3428                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3429                         $this->mOutput->setNewSection( true );
3430                 }
3431
3432                 # Allow user to remove the "new section"
3433                 # link via __NONEWSECTIONLINK__
3434                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3435                         $this->mOutput->hideNewSection( true );
3436                 }
3437
3438                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3439                 # override above conditions and always show TOC above first header
3440                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3441                         $this->mShowToc = true;
3442                         $enoughToc = true;
3443                 }
3444
3445                 # We need this to perform operations on the HTML
3446                 $sk = $this->mOptions->getSkin();
3447
3448                 # headline counter
3449                 $headlineCount = 0;
3450                 $numVisible = 0;
3451
3452                 # Ugh .. the TOC should have neat indentation levels which can be
3453                 # passed to the skin functions. These are determined here
3454                 $toc = '';
3455                 $full = '';
3456                 $head = array();
3457                 $sublevelCount = array();
3458                 $levelCount = array();
3459                 $toclevel = 0;
3460                 $level = 0;
3461                 $prevlevel = 0;
3462                 $toclevel = 0;
3463                 $prevtoclevel = 0;
3464                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3465                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3466                 $oldType = $this->mOutputType;
3467                 $this->setOutputType( self::OT_WIKI );
3468                 $frame = $this->getPreprocessor()->newFrame();
3469                 $root = $this->preprocessToDom( $origText );
3470                 $node = $root->getFirstChild();
3471                 $byteOffset = 0;
3472                 $tocraw = array();
3473
3474                 foreach( $matches[3] as $headline ) {
3475                         $isTemplate = false;
3476                         $titleText = false;
3477                         $sectionIndex = false;
3478                         $numbering = '';
3479                         $markerMatches = array();
3480                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3481                                 $serial = $markerMatches[1];
3482                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3483                                 $isTemplate = ($titleText != $baseTitleText);
3484                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3485                         }
3486
3487                         if( $toclevel ) {
3488                                 $prevlevel = $level;
3489                                 $prevtoclevel = $toclevel;
3490                         }
3491                         $level = $matches[1][$headlineCount];
3492
3493                         if( $doNumberHeadings || $enoughToc ) {
3494
3495                                 if ( $level > $prevlevel ) {
3496                                         # Increase TOC level
3497                                         $toclevel++;
3498                                         $sublevelCount[$toclevel] = 0;
3499                                         if( $toclevel<$wgMaxTocLevel ) {
3500                                                 $prevtoclevel = $toclevel;
3501                                                 $toc .= $sk->tocIndent();
3502                                                 $numVisible++;
3503                                         }
3504                                 }
3505                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3506                                         # Decrease TOC level, find level to jump to
3507
3508                                         for ($i = $toclevel; $i > 0; $i--) {
3509                                                 if ( $levelCount[$i] == $level ) {
3510                                                         # Found last matching level
3511                                                         $toclevel = $i;
3512                                                         break;
3513                                                 }
3514                                                 elseif ( $levelCount[$i] < $level ) {
3515                                                         # Found first matching level below current level
3516                                                         $toclevel = $i + 1;
3517                                                         break;
3518                                                 }
3519                                         }
3520                                         if( $i == 0 ) $toclevel = 1;
3521                                         if( $toclevel<$wgMaxTocLevel ) {
3522                                                 if($prevtoclevel < $wgMaxTocLevel) {
3523                                                         # Unindent only if the previous toc level was shown :p
3524                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3525                                                         $prevtoclevel = $toclevel;
3526                                                 } else {
3527                                                         $toc .= $sk->tocLineEnd();
3528                                                 }
3529                                         }
3530                                 }
3531                                 else {
3532                                         # No change in level, end TOC line
3533                                         if( $toclevel<$wgMaxTocLevel ) {
3534                                                 $toc .= $sk->tocLineEnd();
3535                                         }
3536                                 }
3537
3538                                 $levelCount[$toclevel] = $level;
3539
3540                                 # count number of headlines for each level
3541                                 @$sublevelCount[$toclevel]++;
3542                                 $dot = 0;
3543                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3544                                         if( !empty( $sublevelCount[$i] ) ) {
3545                                                 if( $dot ) {
3546                                                         $numbering .= '.';
3547                                                 }
3548                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3549                                                 $dot = 1;
3550                                         }
3551                                 }
3552                         }
3553
3554                         # The safe header is a version of the header text safe to use for links
3555                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3556                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3557
3558                         # Remove link placeholders by the link text.
3559                         #     <!--LINK number-->
3560                         # turns into
3561                         #     link text with suffix
3562                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3563
3564                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3565                         $tocline = preg_replace(
3566                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3567                                 array( '',                          '<$1>'),
3568                                 $safeHeadline
3569                         );
3570                         $tocline = trim( $tocline );
3571
3572                         # For the anchor, strip out HTML-y stuff period
3573                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3574                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3575                         $safeHeadline = trim( $safeHeadline );
3576
3577                         # Save headline for section edit hint before it's escaped
3578                         $headlineHint = $safeHeadline;
3579
3580                         if ( $wgEnforceHtmlIds ) {
3581                                 $legacyHeadline = false;
3582                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3583                                         'noninitial' );
3584                         } else {
3585                                 # For reverse compatibility, provide an id that's
3586                                 # HTML4-compatible, like we used to.
3587                                 #
3588                                 # It may be worth noting, academically, that it's possible for
3589                                 # the legacy anchor to conflict with a non-legacy headline
3590                                 # anchor on the page.  In this case likely the "correct" thing
3591                                 # would be to either drop the legacy anchors or make sure
3592                                 # they're numbered first.  However, this would require people
3593                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3594                                 # manually, so let's not bother worrying about it.
3595                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3596                                         'noninitial' );
3597                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3598
3599                                 if ( $legacyHeadline == $safeHeadline ) {
3600                                         # No reason to have both (in fact, we can't)
3601                                         $legacyHeadline = false;
3602                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3603                                 $legacyHeadline, 'xml' ) ) {
3604                                         # The legacy id is invalid XML.  We used to allow this, but
3605                                         # there's no reason to do so anymore.  Backward
3606                                         # compatibility will fail slightly in this case, but it's
3607                                         # no big deal.
3608                                         $legacyHeadline = false;
3609                                 }
3610                         }
3611
3612                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3613                         # Does this apply to Unicode characters?  Because we aren't
3614                         # handling those here.
3615                         $arrayKey = strtolower( $safeHeadline );
3616                         if ( $legacyHeadline === false ) {
3617                                 $legacyArrayKey = false;
3618                         } else {
3619                                 $legacyArrayKey = strtolower( $legacyHeadline );
3620                         }
3621
3622                         # count how many in assoc. array so we can track dupes in anchors
3623                         if ( isset( $refers[$arrayKey] ) ) {
3624                                 $refers[$arrayKey]++;
3625                         } else {
3626                                 $refers[$arrayKey] = 1;
3627                         }
3628                         if ( isset( $refers[$legacyArrayKey] ) ) {
3629                                 $refers[$legacyArrayKey]++;
3630                         } else {
3631                                 $refers[$legacyArrayKey] = 1;
3632                         }
3633
3634                         # Don't number the heading if it is the only one (looks silly)
3635                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3636                                 # the two are different if the line contains a link
3637                                 $headline=$numbering . ' ' . $headline;
3638                         }
3639
3640                         # Create the anchor for linking from the TOC to the section
3641                         $anchor = $safeHeadline;
3642                         $legacyAnchor = $legacyHeadline;
3643                         if ( $refers[$arrayKey] > 1 ) {
3644                                 $anchor .= '_' . $refers[$arrayKey];
3645                         }
3646                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3647                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3648                         }
3649                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3650                                 $toc .= $sk->tocLine($anchor, $tocline,
3651                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3652
3653                                 # Find the DOM node for this header
3654                                 while ( $node && !$isTemplate ) {
3655                                         if ( $node->getName() === 'h' ) {
3656                                                 $bits = $node->splitHeading();
3657                                                 if ( $bits['i'] == $sectionIndex )
3658                                                         break;
3659                                         }
3660                                         $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3661                                                 $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3662                                         $node = $node->getNextSibling();
3663                                 }
3664                                 $tocraw[] = array(
3665                                         'toclevel' => $toclevel,
3666                                         'level' => $level,
3667                                         'line' => $tocline,
3668                                         'number' => $numbering,
3669                                         'index' => $sectionIndex,
3670                                         'fromtitle' => $titleText,
3671                                         'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3672                                 );
3673                         }
3674                         # give headline the correct <h#> tag
3675                         if( $showEditLink && $sectionIndex !== false ) {
3676                                 if( $isTemplate ) {
3677                                         # Put a T flag in the section identifier, to indicate to extractSections()
3678                                         # that sections inside <includeonly> should be counted.
3679                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3680                                 } else {
3681                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3682                                 }
3683                         } else {
3684                                 $editlink = '';
3685                         }
3686                         $head[$headlineCount] = $sk->makeHeadline( $level,
3687                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3688                                 $editlink, $legacyAnchor );
3689
3690                         $headlineCount++;
3691                 }
3692
3693                 $this->setOutputType( $oldType );
3694                 $this->mOutput->setSections( $tocraw );
3695
3696                 # Never ever show TOC if no headers
3697                 if( $numVisible < 1 ) {
3698                         $enoughToc = false;
3699                 }
3700
3701                 if( $enoughToc ) {
3702                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3703                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3704                         }
3705                         $toc = $sk->tocList( $toc );
3706                 }
3707
3708                 $this->mOutput->setTOCHTML( $toc );
3709
3710                 # split up and insert constructed headlines
3711
3712                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3713                 $i = 0;
3714
3715                 foreach( $blocks as $block ) {
3716                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3717                                 # This is the [edit] link that appears for the top block of text when
3718                                 # section editing is enabled
3719
3720                                 # Disabled because it broke block formatting
3721                                 # For example, a bullet point in the top line
3722                                 # $full .= $sk->editSectionLink(0);
3723                         }
3724                         $full .= $block;
3725                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3726                                 # Top anchor now in skin
3727                                 $full = $full.$toc;
3728                         }
3729
3730                         if( !empty( $head[$i] ) ) {
3731                                 $full .= $head[$i];
3732                         }
3733                         $i++;
3734                 }
3735                 if( $this->mForceTocPosition ) {
3736                         return str_replace( '<!--MWTOC-->', $toc, $full );
3737                 } else {
3738                         return $full;
3739                 }
3740         }
3741
3742         /**
3743          * Transform wiki markup when saving a page by doing \r\n -> \n
3744          * conversion, substitting signatures, {{subst:}} templates, etc.
3745          *
3746          * @param string $text the text to transform
3747          * @param Title &$title the Title object for the current article
3748          * @param User $user the User object describing the current user
3749          * @param ParserOptions $options parsing options
3750          * @param bool $clearState whether to clear the parser state first
3751          * @return string the altered wiki markup
3752          * @public
3753          */
3754         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3755                 $this->mOptions = $options;
3756                 $this->setTitle( $title );
3757                 $this->setOutputType( self::OT_WIKI );
3758
3759                 if ( $clearState ) {
3760                         $this->clearState();
3761                 }
3762
3763                 $pairs = array(
3764                         "\r\n" => "\n",
3765                 );
3766                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3767                 $text = $this->pstPass2( $text, $user );
3768                 $text = $this->mStripState->unstripBoth( $text );
3769                 return $text;
3770         }
3771
3772         /**
3773          * Pre-save transform helper function
3774          * @private
3775          */
3776         function pstPass2( $text, $user ) {
3777                 global $wgContLang, $wgLocaltimezone;
3778
3779                 /* Note: This is the timestamp saved as hardcoded wikitext to
3780                  * the database, we use $wgContLang here in order to give
3781                  * everyone the same signature and use the default one rather
3782                  * than the one selected in each user's preferences.
3783                  *
3784                  * (see also bug 12815)
3785                  */
3786                 $ts = $this->mOptions->getTimestamp();
3787                 $tz = wfMsgForContent( 'timezone-utc' );
3788                 if ( isset( $wgLocaltimezone ) ) {
3789                         $unixts = wfTimestamp( TS_UNIX, $ts );
3790                         $oldtz = getenv( 'TZ' );
3791                         putenv( 'TZ='.$wgLocaltimezone );
3792                         $ts = date( 'YmdHis', $unixts );
3793                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3794
3795                         /* Allow translation of timezones trough wiki. date() can return
3796                          * whatever crap the system uses, localised or not, so we cannot
3797                          * ship premade translations.
3798                          */
3799                         $key = 'timezone-' . strtolower( trim( $tz ) );
3800                         $value = wfMsgForContent( $key );
3801                         if ( !wfEmptyMsg( $key, $value ) ) $tz = $value;
3802
3803                         putenv( 'TZ='.$oldtz );
3804                 }
3805
3806                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3807
3808                 # Variable replacement
3809                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3810                 $text = $this->replaceVariables( $text );
3811
3812                 # Signatures
3813                 $sigText = $this->getUserSig( $user );
3814                 $text = strtr( $text, array(
3815                         '~~~~~' => $d,
3816                         '~~~~' => "$sigText $d",
3817                         '~~~' => $sigText
3818                 ) );
3819
3820                 # Context links: [[|name]] and [[name (context)|]]
3821                 #
3822                 global $wgLegalTitleChars;
3823                 $tc = "[$wgLegalTitleChars]";
3824                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3825
3826                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3827                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3828                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3829                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3830
3831                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3832                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3833                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3834                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3835
3836                 $t = $this->mTitle->getText();
3837                 $m = array();
3838                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3839                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3840                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3841                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3842                 } else {
3843                         # if there's no context, don't bother duplicating the title
3844                         $text = preg_replace( $p2, '[[\\1]]', $text );
3845                 }
3846
3847                 # Trim trailing whitespace
3848                 $text = rtrim( $text );
3849
3850                 return $text;
3851         }
3852
3853         /**
3854          * Fetch the user's signature text, if any, and normalize to
3855          * validated, ready-to-insert wikitext.
3856          *
3857          * @param User $user
3858          * @return string
3859          * @private
3860          */
3861         function getUserSig( &$user ) {
3862                 global $wgMaxSigChars;
3863
3864                 $username = $user->getName();
3865                 $nickname = $user->getOption( 'nickname' );
3866                 $nickname = $nickname === null ? $username : $nickname;
3867
3868                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3869                         $nickname = $username;
3870                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3871                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3872                         # Sig. might contain markup; validate this
3873                         if( $this->validateSig( $nickname ) !== false ) {
3874                                 # Validated; clean up (if needed) and return it
3875                                 return $this->cleanSig( $nickname, true );
3876                         } else {
3877                                 # Failed to validate; fall back to the default
3878                                 $nickname = $username;
3879                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3880                         }
3881                 }
3882
3883                 // Make sure nickname doesnt get a sig in a sig
3884                 $nickname = $this->cleanSigInSig( $nickname );
3885
3886                 # If we're still here, make it a link to the user page
3887                 $userText = wfEscapeWikiText( $username );
3888                 $nickText = wfEscapeWikiText( $nickname );
3889                 if ( $user->isAnon() )  {
3890                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3891                 } else {
3892                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3893                 }
3894         }
3895
3896         /**
3897          * Check that the user's signature contains no bad XML
3898          *
3899          * @param string $text
3900          * @return mixed An expanded string, or false if invalid.
3901          */
3902         function validateSig( $text ) {
3903                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3904         }
3905
3906         /**
3907          * Clean up signature text
3908          *
3909          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3910          * 2) Substitute all transclusions
3911          *
3912          * @param string $text
3913          * @param $parsing Whether we're cleaning (preferences save) or parsing
3914          * @return string Signature text
3915          */
3916         function cleanSig( $text, $parsing = false ) {
3917                 if ( !$parsing ) {
3918                         global $wgTitle;
3919                         $this->clearState();
3920                         $this->setTitle( $wgTitle );
3921                         $this->mOptions = new ParserOptions;
3922                         $this->setOutputType = self::OT_PREPROCESS;
3923                 }
3924
3925                 # Option to disable this feature
3926                 if ( !$this->mOptions->getCleanSignatures() ) {
3927                         return $text;
3928                 }
3929
3930                 # FIXME: regex doesn't respect extension tags or nowiki
3931                 #  => Move this logic to braceSubstitution()
3932                 $substWord = MagicWord::get( 'subst' );
3933                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3934                 $substText = '{{' . $substWord->getSynonym( 0 );
3935
3936                 $text = preg_replace( $substRegex, $substText, $text );
3937                 $text = $this->cleanSigInSig( $text );
3938                 $dom = $this->preprocessToDom( $text );
3939                 $frame = $this->getPreprocessor()->newFrame();
3940                 $text = $frame->expand( $dom );
3941
3942                 if ( !$parsing ) {
3943                         $text = $this->mStripState->unstripBoth( $text );
3944                 }
3945
3946                 return $text;
3947         }
3948
3949         /**
3950          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3951          * @param string $text
3952          * @return string Signature text with /~{3,5}/ removed
3953          */
3954         function cleanSigInSig( $text ) {
3955                 $text = preg_replace( '/~{3,5}/', '', $text );
3956                 return $text;
3957         }
3958
3959         /**
3960          * Set up some variables which are usually set up in parse()
3961          * so that an external function can call some class members with confidence
3962          * @public
3963          */
3964         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3965                 $this->setTitle( $title );
3966                 $this->mOptions = $options;
3967                 $this->setOutputType( $outputType );
3968                 if ( $clearState ) {
3969                         $this->clearState();
3970                 }
3971         }
3972
3973         /**
3974          * Wrapper for preprocess()
3975          *
3976          * @param string $text the text to preprocess
3977          * @param ParserOptions $options  options
3978          * @return string
3979          * @public
3980          */
3981         function transformMsg( $text, $options ) {
3982                 global $wgTitle;
3983                 static $executing = false;
3984
3985                 # Guard against infinite recursion
3986                 if ( $executing ) {
3987                         return $text;
3988                 }
3989                 $executing = true;
3990
3991                 wfProfileIn(__METHOD__);
3992                 $text = $this->preprocess( $text, $wgTitle, $options );
3993
3994                 $executing = false;
3995                 wfProfileOut(__METHOD__);
3996                 return $text;
3997         }
3998
3999         /**
4000          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4001          * The callback should have the following form:
4002          *    function myParserHook( $text, $params, &$parser ) { ... }
4003          *
4004          * Transform and return $text. Use $parser for any required context, e.g. use
4005          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4006          *
4007          * @public
4008          *
4009          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4010          * @param mixed $callback The callback function (and object) to use for the tag
4011          *
4012          * @return The old value of the mTagHooks array associated with the hook
4013          */
4014         function setHook( $tag, $callback ) {
4015                 $tag = strtolower( $tag );
4016                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4017                 $this->mTagHooks[$tag] = $callback;
4018                 if( !in_array( $tag, $this->mStripList ) ) {
4019                         $this->mStripList[] = $tag;
4020                 }
4021
4022                 return $oldVal;
4023         }
4024
4025         function setTransparentTagHook( $tag, $callback ) {
4026                 $tag = strtolower( $tag );
4027                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4028                 $this->mTransparentTagHooks[$tag] = $callback;
4029
4030                 return $oldVal;
4031         }
4032
4033         /**
4034          * Remove all tag hooks
4035          */
4036         function clearTagHooks() {
4037                 $this->mTagHooks = array();
4038                 $this->mStripList = $this->mDefaultStripList;
4039         }
4040
4041         /**
4042          * Create a function, e.g. {{sum:1|2|3}}
4043          * The callback function should have the form:
4044          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4045          *
4046          * Or with SFH_OBJECT_ARGS:
4047          *    function myParserFunction( $parser, $frame, $args ) { ... }
4048          *
4049          * The callback may either return the text result of the function, or an array with the text
4050          * in element 0, and a number of flags in the other elements. The names of the flags are
4051          * specified in the keys. Valid flags are:
4052          *   found                     The text returned is valid, stop processing the template. This
4053          *                             is on by default.
4054          *   nowiki                    Wiki markup in the return value should be escaped
4055          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4056          *
4057          * @public
4058          *
4059          * @param string $id The magic word ID
4060          * @param mixed $callback The callback function (and object) to use
4061          * @param integer $flags a combination of the following flags:
4062          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4063          *
4064          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4065          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4066          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4067          *     the arguments, and to control the way they are expanded.
4068          *
4069          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4070          *     arguments, for instance:
4071          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4072          *
4073          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4074          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4075          *     working if/when this is changed.
4076          *
4077          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4078          *     expansion.
4079          *
4080          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4081          *     about the methods available in PPFrame and PPNode.
4082          *
4083          * @return The old callback function for this name, if any
4084          */
4085         function setFunctionHook( $id, $callback, $flags = 0 ) {
4086                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4087                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4088
4089                 # Add to function cache
4090                 $mw = MagicWord::get( $id );
4091                 if( !$mw )
4092                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4093
4094                 $synonyms = $mw->getSynonyms();
4095                 $sensitive = intval( $mw->isCaseSensitive() );
4096
4097                 foreach ( $synonyms as $syn ) {
4098                         # Case
4099                         if ( !$sensitive ) {
4100                                 $syn = strtolower( $syn );
4101                         }
4102                         # Add leading hash
4103                         if ( !( $flags & SFH_NO_HASH ) ) {
4104                                 $syn = '#' . $syn;
4105                         }
4106                         # Remove trailing colon
4107                         if ( substr( $syn, -1, 1 ) === ':' ) {
4108                                 $syn = substr( $syn, 0, -1 );
4109                         }
4110                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4111                 }
4112                 return $oldVal;
4113         }
4114
4115         /**
4116          * Get all registered function hook identifiers
4117          *
4118          * @return array
4119          */
4120         function getFunctionHooks() {
4121                 return array_keys( $this->mFunctionHooks );
4122         }
4123
4124         /**
4125          * FIXME: update documentation. makeLinkObj() is deprecated.
4126          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4127          * Placeholders created in Skin::makeLinkObj()
4128          * Returns an array of link CSS classes, indexed by PDBK.
4129          */
4130         function replaceLinkHolders( &$text, $options = 0 ) {
4131                 return $this->mLinkHolders->replace( $text );
4132         }
4133
4134         /**
4135          * Replace <!--LINK--> link placeholders with plain text of links
4136          * (not HTML-formatted).
4137          * @param string $text
4138          * @return string
4139          */
4140         function replaceLinkHoldersText( $text ) {
4141                 return $this->mLinkHolders->replaceText( $text );
4142         }
4143
4144         /**
4145          * Tag hook handler for 'pre'.
4146          */
4147         function renderPreTag( $text, $attribs ) {
4148                 // Backwards-compatibility hack
4149                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4150
4151                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4152                 return Xml::openElement( 'pre', $attribs ) .
4153                         Xml::escapeTagsOnly( $content ) .
4154                         '</pre>';
4155         }
4156
4157         /**
4158          * Renders an image gallery from a text with one line per image.
4159          * text labels may be given by using |-style alternative text. E.g.
4160          *   Image:one.jpg|The number "1"
4161          *   Image:tree.jpg|A tree
4162          * given as text will return the HTML of a gallery with two images,
4163          * labeled 'The number "1"' and
4164          * 'A tree'.
4165          */
4166         function renderImageGallery( $text, $params ) {
4167                 $ig = new ImageGallery();
4168                 $ig->setContextTitle( $this->mTitle );
4169                 $ig->setShowBytes( false );
4170                 $ig->setShowFilename( false );
4171                 $ig->setParser( $this );
4172                 $ig->setHideBadImages();
4173                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4174                 $ig->useSkin( $this->mOptions->getSkin() );
4175                 $ig->mRevisionId = $this->mRevisionId;
4176
4177                 if( isset( $params['caption'] ) ) {
4178                         $caption = $params['caption'];
4179                         $caption = htmlspecialchars( $caption );
4180                         $caption = $this->replaceInternalLinks( $caption );
4181                         $ig->setCaptionHtml( $caption );
4182                 }
4183                 if( isset( $params['perrow'] ) ) {
4184                         $ig->setPerRow( $params['perrow'] );
4185                 }
4186                 if( isset( $params['widths'] ) ) {
4187                         $ig->setWidths( $params['widths'] );
4188                 }
4189                 if( isset( $params['heights'] ) ) {
4190                         $ig->setHeights( $params['heights'] );
4191                 }
4192
4193                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4194
4195                 $lines = StringUtils::explode( "\n", $text );
4196                 foreach ( $lines as $line ) {
4197                         # match lines like these:
4198                         # Image:someimage.jpg|This is some image
4199                         $matches = array();
4200                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4201                         # Skip empty lines
4202                         if ( count( $matches ) == 0 ) {
4203                                 continue;
4204                         }
4205
4206                         if ( strpos( $matches[0], '%' ) !== false )
4207                                 $matches[1] = urldecode( $matches[1] );
4208                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4209                         $nt =& $tp;
4210                         if( is_null( $nt ) ) {
4211                                 # Bogus title. Ignore these so we don't bomb out later.
4212                                 continue;
4213                         }
4214                         if ( isset( $matches[3] ) ) {
4215                                 $label = $matches[3];
4216                         } else {
4217                                 $label = '';
4218                         }
4219
4220                         $html = $this->recursiveTagParse( trim( $label ) );
4221
4222                         $ig->add( $nt, $html );
4223
4224                         # Only add real images (bug #5586)
4225                         if ( $nt->getNamespace() == NS_FILE ) {
4226                                 $this->mOutput->addImage( $nt->getDBkey() );
4227                         }
4228                 }
4229                 return $ig->toHTML();
4230         }
4231
4232         function getImageParams( $handler ) {
4233                 if ( $handler ) {
4234                         $handlerClass = get_class( $handler );
4235                 } else {
4236                         $handlerClass = '';
4237                 }
4238                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4239                         // Initialise static lists
4240                         static $internalParamNames = array(
4241                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4242                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4243                                         'bottom', 'text-bottom' ),
4244                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4245                                         'upright', 'border', 'link', 'alt' ),
4246                         );
4247                         static $internalParamMap;
4248                         if ( !$internalParamMap ) {
4249                                 $internalParamMap = array();
4250                                 foreach ( $internalParamNames as $type => $names ) {
4251                                         foreach ( $names as $name ) {
4252                                                 $magicName = str_replace( '-', '_', "img_$name" );
4253                                                 $internalParamMap[$magicName] = array( $type, $name );
4254                                         }
4255                                 }
4256                         }
4257
4258                         // Add handler params
4259                         $paramMap = $internalParamMap;
4260                         if ( $handler ) {
4261                                 $handlerParamMap = $handler->getParamMap();
4262                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4263                                         $paramMap[$magic] = array( 'handler', $paramName );
4264                                 }
4265                         }
4266                         $this->mImageParams[$handlerClass] = $paramMap;
4267                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4268                 }
4269                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4270         }
4271
4272         /**
4273          * Parse image options text and use it to make an image
4274          * @param Title $title
4275          * @param string $options
4276          * @param LinkHolderArray $holders
4277          */
4278         function makeImage( $title, $options, $holders = false ) {
4279                 # Check if the options text is of the form "options|alt text"
4280                 # Options are:
4281                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4282                 #  * left       no resizing, just left align. label is used for alt= only
4283                 #  * right      same, but right aligned
4284                 #  * none       same, but not aligned
4285                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4286                 #  * center     center the image
4287                 #  * framed     Keep original image size, no magnify-button.
4288                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4289                 #  * upright    reduce width for upright images, rounded to full __0 px
4290                 #  * border     draw a 1px border around the image
4291                 #  * alt        Text for HTML alt attribute (defaults to empty)
4292                 # vertical-align values (no % or length right now):
4293                 #  * baseline
4294                 #  * sub
4295                 #  * super
4296                 #  * top
4297                 #  * text-top
4298                 #  * middle
4299                 #  * bottom
4300                 #  * text-bottom
4301
4302                 $parts = StringUtils::explode( "|", $options );
4303                 $sk = $this->mOptions->getSkin();
4304
4305                 # Give extensions a chance to select the file revision for us
4306                 $skip = $time = $descQuery = false;
4307                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4308
4309                 if ( $skip ) {
4310                         return $sk->link( $title );
4311                 }
4312
4313                 # Get the file
4314                 $imagename = $title->getDBkey();
4315                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4316                         $file = $this->mFileCache[$imagename][$time];
4317                 } else {
4318                         $file = wfFindFile( $title, $time );
4319                         if ( count( $this->mFileCache ) > 1000 ) {
4320                                 $this->mFileCache = array();
4321                         }
4322                         $this->mFileCache[$imagename][$time] = $file;
4323                 }
4324                 # Get parameter map
4325                 $handler = $file ? $file->getHandler() : false;
4326
4327                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4328
4329                 # Process the input parameters
4330                 $caption = '';
4331                 $params = array( 'frame' => array(), 'handler' => array(),
4332                         'horizAlign' => array(), 'vertAlign' => array() );
4333                 foreach( $parts as $part ) {
4334                         $part = trim( $part );
4335                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4336                         $validated = false;
4337                         if( isset( $paramMap[$magicName] ) ) {
4338                                 list( $type, $paramName ) = $paramMap[$magicName];
4339
4340                                 // Special case; width and height come in one variable together
4341                                 if( $type === 'handler' && $paramName === 'width' ) {
4342                                         $m = array();
4343                                         # (bug 13500) In both cases (width/height and width only),
4344                                         # permit trailing "px" for backward compatibility.
4345                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4346                                                 $width = intval( $m[1] );
4347                                                 $height = intval( $m[2] );
4348                                                 if ( $handler->validateParam( 'width', $width ) ) {
4349                                                         $params[$type]['width'] = $width;
4350                                                         $validated = true;
4351                                                 }
4352                                                 if ( $handler->validateParam( 'height', $height ) ) {
4353                                                         $params[$type]['height'] = $height;
4354                                                         $validated = true;
4355                                                 }
4356                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4357                                                 $width = intval( $value );
4358                                                 if ( $handler->validateParam( 'width', $width ) ) {
4359                                                         $params[$type]['width'] = $width;
4360                                                         $validated = true;
4361                                                 }
4362                                         } // else no validation -- bug 13436
4363                                 } else {
4364                                         if ( $type === 'handler' ) {
4365                                                 # Validate handler parameter
4366                                                 $validated = $handler->validateParam( $paramName, $value );
4367                                         } else {
4368                                                 # Validate internal parameters
4369                                                 switch( $paramName ) {
4370                                                 case 'manualthumb':
4371                                                 case 'alt':
4372                                                         // @fixme - possibly check validity here for
4373                                                         // manualthumb? downstream behavior seems odd with
4374                                                         // missing manual thumbs.
4375                                                         $validated = true;
4376                                                         $value = $this->stripAltText( $value, $holders );
4377                                                         break;
4378                                                 case 'link':
4379                                                         $chars = self::EXT_LINK_URL_CLASS;
4380                                                         $prots = $this->mUrlProtocols;
4381                                                         if ( $value === '' ) {
4382                                                                 $paramName = 'no-link';
4383                                                                 $value = true;
4384                                                                 $validated = true;
4385                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4386                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4387                                                                         $paramName = 'link-url';
4388                                                                         $this->mOutput->addExternalLink( $value );
4389                                                                         $validated = true;
4390                                                                 }
4391                                                         } else {
4392                                                                 $linkTitle = Title::newFromText( $value );
4393                                                                 if ( $linkTitle ) {
4394                                                                         $paramName = 'link-title';
4395                                                                         $value = $linkTitle;
4396                                                                         $this->mOutput->addLink( $linkTitle );
4397                                                                         $validated = true;
4398                                                                 }
4399                                                         }
4400                                                         break;
4401                                                 default:
4402                                                         // Most other things appear to be empty or numeric...
4403                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4404                                                 }
4405                                         }
4406
4407                                         if ( $validated ) {
4408                                                 $params[$type][$paramName] = $value;
4409                                         }
4410                                 }
4411                         }
4412                         if ( !$validated ) {
4413                                 $caption = $part;
4414                         }
4415                 }
4416
4417                 # Process alignment parameters
4418                 if ( $params['horizAlign'] ) {
4419                         $params['frame']['align'] = key( $params['horizAlign'] );
4420                 }
4421                 if ( $params['vertAlign'] ) {
4422                         $params['frame']['valign'] = key( $params['vertAlign'] );
4423                 }
4424
4425                 $params['frame']['caption'] = $caption;
4426
4427                 $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4428
4429                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4430                 # came to also set the caption, ordinary text after the image -- which
4431                 # makes no sense, because that just repeats the text multiple times in
4432                 # screen readers.  It *also* came to set the title attribute.
4433                 #
4434                 # Now that we have an alt attribute, we should not set the alt text to
4435                 # equal the caption: that's worse than useless, it just repeats the
4436                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4437                 # use the unnamed parameter for alt text as well, just for the time be-
4438                 # ing, if the unnamed param is set and the alt param is not.
4439                 #
4440                 # For the future, we need to figure out if we want to tweak this more,
4441                 # e.g., introducing a title= parameter for the title; ignoring the un-
4442                 # named parameter entirely for images without a caption; adding an ex-
4443                 # plicit caption= parameter and preserving the old magic unnamed para-
4444                 # meter for BC; ...
4445                 if( $caption !== '' && !isset( $params['frame']['alt'] )
4446                 && !isset( $params['frame']['framed'] )
4447                 && !isset( $params['frame']['thumbnail'] )
4448                 && !isset( $params['frame']['manualthumb'] ) ) {
4449                         $params['frame']['alt'] = $params['frame']['title'];
4450                 }
4451
4452                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4453
4454                 # Linker does the rest
4455                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4456
4457                 # Give the handler a chance to modify the parser object
4458                 if ( $handler ) {
4459                         $handler->parserTransformHook( $this, $file );
4460                 }
4461
4462                 return $ret;
4463         }
4464
4465         protected function stripAltText( $caption, $holders ) {
4466                 # Strip bad stuff out of the title (tooltip).  We can't just use
4467                 # replaceLinkHoldersText() here, because if this function is called
4468                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4469                 if ( $holders ) {
4470                         $tooltip = $holders->replaceText( $caption );
4471                 } else {
4472                         $tooltip = $this->replaceLinkHoldersText( $caption );
4473                 }
4474
4475                 # make sure there are no placeholders in thumbnail attributes
4476                 # that are later expanded to html- so expand them now and
4477                 # remove the tags
4478                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4479                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4480
4481                 return $tooltip;
4482         }
4483
4484         /**
4485          * Set a flag in the output object indicating that the content is dynamic and
4486          * shouldn't be cached.
4487          */
4488         function disableCache() {
4489                 wfDebug( "Parser output marked as uncacheable.\n" );
4490                 $this->mOutput->mCacheTime = -1;
4491         }
4492
4493         /**#@+
4494          * Callback from the Sanitizer for expanding items found in HTML attribute
4495          * values, so they can be safely tested and escaped.
4496          * @param string $text
4497          * @param PPFrame $frame
4498          * @return string
4499          * @private
4500          */
4501         function attributeStripCallback( &$text, $frame = false ) {
4502                 $text = $this->replaceVariables( $text, $frame );
4503                 $text = $this->mStripState->unstripBoth( $text );
4504                 return $text;
4505         }
4506
4507         /**#@-*/
4508
4509         /**#@+
4510          * Accessor/mutator
4511          */
4512         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4513         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4514         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4515         /**#@-*/
4516
4517         /**#@+
4518          * Accessor
4519          */
4520         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4521         /**#@-*/
4522
4523
4524         /**
4525          * Break wikitext input into sections, and either pull or replace
4526          * some particular section's text.
4527          *
4528          * External callers should use the getSection and replaceSection methods.
4529          *
4530          * @param string $text Page wikitext
4531          * @param string $section A section identifier string of the form:
4532          *   <flag1> - <flag2> - ... - <section number>
4533          *
4534          * Currently the only recognised flag is "T", which means the target section number
4535          * was derived during a template inclusion parse, in other words this is a template
4536          * section edit link. If no flags are given, it was an ordinary section edit link.
4537          * This flag is required to avoid a section numbering mismatch when a section is
4538          * enclosed by <includeonly> (bug 6563).
4539          *
4540          * The section number 0 pulls the text before the first heading; other numbers will
4541          * pull the given section along with its lower-level subsections. If the section is
4542          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4543          *
4544          * @param string $mode One of "get" or "replace"
4545          * @param string $newText Replacement text for section data.
4546          * @return string for "get", the extracted section text.
4547          *                for "replace", the whole page with the section replaced.
4548          */
4549         private function extractSections( $text, $section, $mode, $newText='' ) {
4550                 global $wgTitle;
4551                 $this->clearState();
4552                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4553                 $this->mOptions = new ParserOptions;
4554                 $this->setOutputType( self::OT_WIKI );
4555                 $outText = '';
4556                 $frame = $this->getPreprocessor()->newFrame();
4557
4558                 // Process section extraction flags
4559                 $flags = 0;
4560                 $sectionParts = explode( '-', $section );
4561                 $sectionIndex = array_pop( $sectionParts );
4562                 foreach ( $sectionParts as $part ) {
4563                         if ( $part === 'T' ) {
4564                                 $flags |= self::PTD_FOR_INCLUSION;
4565                         }
4566                 }
4567                 // Preprocess the text
4568                 $root = $this->preprocessToDom( $text, $flags );
4569
4570                 // <h> nodes indicate section breaks
4571                 // They can only occur at the top level, so we can find them by iterating the root's children
4572                 $node = $root->getFirstChild();
4573
4574                 // Find the target section
4575                 if ( $sectionIndex == 0 ) {
4576                         // Section zero doesn't nest, level=big
4577                         $targetLevel = 1000;
4578                 } else {
4579             while ( $node ) {
4580                 if ( $node->getName() === 'h' ) {
4581                     $bits = $node->splitHeading();
4582                                         if ( $bits['i'] == $sectionIndex ) {
4583                                         $targetLevel = $bits['level'];
4584                                                 break;
4585                                         }
4586                                 }
4587                                 if ( $mode === 'replace' ) {
4588                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4589                                 }
4590                                 $node = $node->getNextSibling();
4591                         }
4592                 }
4593
4594                 if ( !$node ) {
4595                         // Not found
4596                         if ( $mode === 'get' ) {
4597                                 return $newText;
4598                         } else {
4599                                 return $text;
4600                         }
4601                 }
4602
4603                 // Find the end of the section, including nested sections
4604                 do {
4605                         if ( $node->getName() === 'h' ) {
4606                                 $bits = $node->splitHeading();
4607                                 $curLevel = $bits['level'];
4608                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4609                                         break;
4610                                 }
4611                         }
4612                         if ( $mode === 'get' ) {
4613                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4614                         }
4615                         $node = $node->getNextSibling();
4616                 } while ( $node );
4617
4618                 // Write out the remainder (in replace mode only)
4619                 if ( $mode === 'replace' ) {
4620                         // Output the replacement text
4621                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4622                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4623                         // Only add trailing whitespace if there is newText
4624                         if($newText != "") {
4625                                 $outText .= $newText . "\n\n";
4626                         }
4627
4628                         while ( $node ) {
4629                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4630                                 $node = $node->getNextSibling();
4631                         }
4632                 }
4633
4634                 if ( is_string( $outText ) ) {
4635                         // Re-insert stripped tags
4636                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4637                 }
4638
4639                 return $outText;
4640         }
4641
4642         /**
4643          * This function returns the text of a section, specified by a number ($section).
4644          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4645          * the first section before any such heading (section 0).
4646          *
4647          * If a section contains subsections, these are also returned.
4648          *
4649          * @param string $text text to look in
4650          * @param string $section section identifier
4651          * @param string $deftext default to return if section is not found
4652          * @return string text of the requested section
4653          */
4654         public function getSection( $text, $section, $deftext='' ) {
4655                 return $this->extractSections( $text, $section, "get", $deftext );
4656         }
4657
4658         public function replaceSection( $oldtext, $section, $text ) {
4659                 return $this->extractSections( $oldtext, $section, "replace", $text );
4660         }
4661
4662         /**
4663          * Get the timestamp associated with the current revision, adjusted for
4664          * the default server-local timestamp
4665          */
4666         function getRevisionTimestamp() {
4667                 if ( is_null( $this->mRevisionTimestamp ) ) {
4668                         wfProfileIn( __METHOD__ );
4669                         global $wgContLang;
4670                         $dbr = wfGetDB( DB_SLAVE );
4671                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4672                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4673
4674                         // Normalize timestamp to internal MW format for timezone processing.
4675                         // This has the added side-effect of replacing a null value with
4676                         // the current time, which gives us more sensible behavior for
4677                         // previews.
4678                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4679
4680                         // The cryptic '' timezone parameter tells to use the site-default
4681                         // timezone offset instead of the user settings.
4682                         //
4683                         // Since this value will be saved into the parser cache, served
4684                         // to other users, and potentially even used inside links and such,
4685                         // it needs to be consistent for all visitors.
4686                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4687
4688                         wfProfileOut( __METHOD__ );
4689                 }
4690                 return $this->mRevisionTimestamp;
4691         }
4692
4693         /**
4694          * Get the name of the user that edited the last revision
4695          */
4696         function getRevisionUser() {
4697                 // if this template is subst: the revision id will be blank,
4698                 // so just use the current user's name
4699                 if( $this->mRevisionId ) {
4700                         $revision = Revision::newFromId( $this->mRevisionId );
4701                         $revuser = $revision->getUserText();
4702                 } else {
4703                         global $wgUser;
4704                         $revuser = $wgUser->getName();
4705                 }
4706                 return $revuser;
4707         }
4708
4709         /**
4710          * Mutator for $mDefaultSort
4711          *
4712          * @param $sort New value
4713          */
4714         public function setDefaultSort( $sort ) {
4715                 $this->mDefaultSort = $sort;
4716         }
4717
4718         /**
4719          * Accessor for $mDefaultSort
4720          * Will use the title/prefixed title if none is set
4721          *
4722          * @return string
4723          */
4724         public function getDefaultSort() {
4725                 global $wgCategoryPrefixedDefaultSortkey;
4726                 if( $this->mDefaultSort !== false ) {
4727                         return $this->mDefaultSort;
4728                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4729                         !$wgCategoryPrefixedDefaultSortkey) {
4730                         return $this->mTitle->getText();
4731                 } else {
4732                         return $this->mTitle->getPrefixedText();
4733                 }
4734         }
4735
4736         /**
4737          * Accessor for $mDefaultSort
4738          * Unlike getDefaultSort(), will return false if none is set
4739          *
4740          * @return string or false
4741          */
4742         public function getCustomDefaultSort() {
4743                 return $this->mDefaultSort;
4744         }
4745
4746         /**
4747          * Try to guess the section anchor name based on a wikitext fragment
4748          * presumably extracted from a heading, for example "Header" from
4749          * "== Header ==".
4750          */
4751         public function guessSectionNameFromWikiText( $text ) {
4752                 # Strip out wikitext links(they break the anchor)
4753                 $text = $this->stripSectionName( $text );
4754                 $headline = Sanitizer::decodeCharReferences( $text );
4755                 # strip out HTML
4756                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4757                 $headline = trim( $headline );
4758                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4759                 $replacearray = array(
4760                         '%3A' => ':',
4761                         '%' => '.'
4762                 );
4763                 return str_replace(
4764                         array_keys( $replacearray ),
4765                         array_values( $replacearray ),
4766                         $sectionanchor );
4767         }
4768
4769         /**
4770          * Strips a text string of wikitext for use in a section anchor
4771          *
4772          * Accepts a text string and then removes all wikitext from the
4773          * string and leaves only the resultant text (i.e. the result of
4774          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4775          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4776          * to create valid section anchors by mimicing the output of the
4777          * parser when headings are parsed.
4778          *
4779          * @param $text string Text string to be stripped of wikitext
4780          * for use in a Section anchor
4781          * @return Filtered text string
4782          */
4783         public function stripSectionName( $text ) {
4784                 # Strip internal link markup
4785                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4786                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4787
4788                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4789                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4790                 # on how many empty links there are on the page - need to figure that out.
4791                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4792
4793                 # Parse wikitext quotes (italics & bold)
4794                 $text = $this->doQuotes($text);
4795
4796                 # Strip HTML tags
4797                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4798                 return $text;
4799         }
4800
4801         function srvus( $text ) {
4802                 return $this->testSrvus( $text, $this->mOutputType );
4803         }
4804
4805         /**
4806          * strip/replaceVariables/unstrip for preprocessor regression testing
4807          */
4808         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4809                 $this->clearState();
4810                 if ( ! ( $title instanceof Title ) ) {
4811                         $title = Title::newFromText( $title );
4812                 }
4813                 $this->mTitle = $title;
4814                 $this->mOptions = $options;
4815                 $this->setOutputType( $outputType );
4816                 $text = $this->replaceVariables( $text );
4817                 $text = $this->mStripState->unstripBoth( $text );
4818                 $text = Sanitizer::removeHTMLtags( $text );
4819                 return $text;
4820         }
4821
4822         function testPst( $text, $title, $options ) {
4823                 global $wgUser;
4824                 if ( ! ( $title instanceof Title ) ) {
4825                         $title = Title::newFromText( $title );
4826                 }
4827                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4828         }
4829
4830         function testPreprocess( $text, $title, $options ) {
4831                 if ( ! ( $title instanceof Title ) ) {
4832                         $title = Title::newFromText( $title );
4833                 }
4834                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4835         }
4836
4837         function markerSkipCallback( $s, $callback ) {
4838                 $i = 0;
4839                 $out = '';
4840                 while ( $i < strlen( $s ) ) {
4841                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4842                         if ( $markerStart === false ) {
4843                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4844                                 break;
4845                         } else {
4846                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4847                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4848                                 if ( $markerEnd === false ) {
4849                                         $out .= substr( $s, $markerStart );
4850                                         break;
4851                                 } else {
4852                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4853                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4854                                         $i = $markerEnd;
4855                                 }
4856                         }
4857                 }
4858                 return $out;
4859         }
4860
4861         function serialiseHalfParsedText( $text ) {
4862                 $data = array();
4863                 $data['text'] = $text;
4864
4865                 // First, find all strip markers, and store their
4866                 //  data in an array.
4867                 $stripState = new StripState;
4868                 $pos = 0;
4869                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
4870                         $end_pos += strlen( self::MARKER_SUFFIX );
4871                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
4872
4873                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
4874                                 $replaceArray = $stripState->general;
4875                                 $stripText = $this->mStripState->general->data[$marker];
4876                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
4877                                 $replaceArray = $stripState->nowiki;
4878                                 $stripText = $this->mStripState->nowiki->data[$marker];
4879                         } else {
4880                                 throw new MWException( "Hanging strip marker: '$marker'." );
4881                         }
4882
4883                         $replaceArray->setPair( $marker, $stripText );
4884                         $pos = $end_pos;
4885                 }
4886                 $data['stripstate'] = $stripState;
4887
4888                 // Now, find all of our links, and store THEIR
4889                 //  data in an array! :)
4890                 $links = array( 'internal' => array(), 'interwiki' => array() );
4891                 $pos = 0;
4892
4893                 // Internal links
4894                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
4895                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
4896
4897                         $ns = trim($ns);
4898                         if (empty( $links['internal'][$ns] )) {
4899                                 $links['internal'][$ns] = array();
4900                         }
4901
4902                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
4903                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
4904                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
4905                 }
4906
4907                 $pos = 0;
4908
4909                 // Interwiki links
4910                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
4911                         $data = substr( $text, $start_pos );
4912                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
4913                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
4914                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
4915                 }
4916
4917                 $data['linkholder'] = $links;
4918
4919                 return $data;
4920         }
4921
4922         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
4923                 if (!$intPrefix)
4924                         $intPrefix = $this->getRandomString();
4925
4926                 // First, extract the strip state.
4927                 $stripState = $data['stripstate'];
4928                 $this->mStripState->general->merge( $stripState->general );
4929                 $this->mStripState->nowiki->merge( $stripState->nowiki );
4930
4931                 // Now, extract the text, and renumber links
4932                 $text = $data['text'];
4933                 $links = $data['linkholder'];
4934
4935                 // Internal...
4936                 foreach( $links['internal'] as $ns => $nsLinks ) {
4937                         foreach( $nsLinks as $key => $entry ) {
4938                                 $newKey = $intPrefix . '-' . $key;
4939                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
4940
4941                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
4942                         }
4943                 }
4944
4945                 // Interwiki...
4946                 foreach( $links['interwiki'] as $key => $entry ) {
4947                         $newKey = "$intPrefix-$key";
4948                         $this->mLinkHolders->interwikis[$newKey] = $entry;
4949
4950                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
4951                 }
4952
4953                 // Should be good to go.
4954                 return $text;
4955         }
4956 }
4957
4958 /**
4959  * @todo document, briefly.
4960  * @ingroup Parser
4961  */
4962 class StripState {
4963         var $general, $nowiki;
4964
4965         function __construct() {
4966                 $this->general = new ReplacementArray;
4967                 $this->nowiki = new ReplacementArray;
4968         }
4969
4970         function unstripGeneral( $text ) {
4971                 wfProfileIn( __METHOD__ );
4972                 do {
4973                         $oldText = $text;
4974                         $text = $this->general->replace( $text );
4975                 } while ( $text !== $oldText );
4976                 wfProfileOut( __METHOD__ );
4977                 return $text;
4978         }
4979
4980         function unstripNoWiki( $text ) {
4981                 wfProfileIn( __METHOD__ );
4982                 do {
4983                         $oldText = $text;
4984                         $text = $this->nowiki->replace( $text );
4985                 } while ( $text !== $oldText );
4986                 wfProfileOut( __METHOD__ );
4987                 return $text;
4988         }
4989
4990         function unstripBoth( $text ) {
4991                 wfProfileIn( __METHOD__ );
4992                 do {
4993                         $oldText = $text;
4994                         $text = $this->general->replace( $text );
4995                         $text = $this->nowiki->replace( $text );
4996                 } while ( $text !== $oldText );
4997                 wfProfileOut( __METHOD__ );
4998                 return $text;
4999         }
5000 }
5001
5002 /**
5003  * @todo document, briefly.
5004  * @ingroup Parser
5005  */
5006 class OnlyIncludeReplacer {
5007         var $output = '';
5008
5009         function replace( $matches ) {
5010                 if ( substr( $matches[1], -1 ) === "\n" ) {
5011                         $this->output .= substr( $matches[1], 0, -1 );
5012                 } else {
5013                         $this->output .= $matches[1];
5014                 }
5015         }
5016 }