includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   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 four main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * transformMsg()
  24  *   performs brace substitution on MediaWiki messages
  25  * preprocess()
  26  *   removes HTML comments and expands templates
  27  *
  28  * Globals used:
  29  *    objects:   $wgLang, $wgContLang
  30  *
  31  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  32  *
  33  * settings:
  34  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  35  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  36  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  37  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.4';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  62                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  63
  64         // State constants for the definition list colon extraction
  65         const COLON_STATE_TEXT = 0;
  66         const COLON_STATE_TAG = 1;
  67         const COLON_STATE_TAGSTART = 2;
  68         const COLON_STATE_CLOSETAG = 3;
  69         const COLON_STATE_TAGSLASH = 4;
  70         const COLON_STATE_COMMENT = 5;
  71         const COLON_STATE_COMMENTDASH = 6;
  72         const COLON_STATE_COMMENTDASHDASH = 7;
  73
  74         // Flags for preprocessToDom
  75         const PTD_FOR_INCLUSION = 1;
  76
  77         /**#@+
  78          * @private
  79          */
  80         # Persistent:
  81         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  82                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  83                 $mExtLinkBracketedRegex;
  84
  85         # Cleared with clearState():
  86         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  87         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  88         var $mInterwikiLinkHolders, $mLinkHolders;
  89         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  90         var $mTplExpandCache; // empty-frame expansion cache
  91         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  92
  93         # Temporary
  94         # These are variables reset at least once per parse regardless of $clearState
  95         var $mOptions,      // ParserOptions object
  96                 $mTitle,        // Title context, used for self-link rendering and similar things
  97                 $mOutputType,   // Output type, one of the OT_xxx constants
  98                 $ot,            // Shortcut alias, see setOutputType()
  99                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 100                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 101                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 102
 103         /**#@-*/
 104
 105         /**
 106          * Constructor
 107          *
 108          * @public
 109          */
 110         function __construct( $conf = array() ) {
 111                 $this->mTagHooks = array();
 112                 $this->mTransparentTagHooks = array();
 113                 $this->mFunctionHooks = array();
 114                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 115                 $this->mStripList = array( 'nowiki', 'gallery' );
 116                 $this->mMarkerSuffix = "-QINU\x7f";
 117                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 118                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 119                 $this->mFirstCall = true;
 120         }
 121
 122         /**
 123          * Do various kinds of initialisation on the first call of the parser
 124          */
 125         function firstCallInit() {
 126                 if ( !$this->mFirstCall ) {
 127                         return;
 128                 }
 129
 130                 wfProfileIn( __METHOD__ );
 131                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 132
 133                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 134
 135                 # Syntax for arguments (see self::setFunctionHook):
 136                 #  "name for lookup in localized magic words array",
 137                 #  function callback,
 138                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 139                 #    instead of {{#int:...}})
 140                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 141                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 142                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 165                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'tag',              array( 'CoreParserFunctions', 'tagObj'           ), SFH_OBJECT_ARGS );
 168
 169                 if ( $wgAllowDisplayTitle ) {
 170                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 171                 }
 172                 if ( $wgAllowSlowParserFunctions ) {
 173                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 174                 }
 175
 176                 $this->initialiseVariables();
 177                 $this->mFirstCall = false;
 178                 wfProfileOut( __METHOD__ );
 179         }
 180
 181         /**
 182          * Clear Parser state
 183          *
 184          * @private
 185          */
 186         function clearState() {
 187                 wfProfileIn( __METHOD__ );
 188                 if ( $this->mFirstCall ) {
 189                         $this->firstCallInit();
 190                 }
 191                 $this->mOutput = new ParserOutput;
 192                 $this->mAutonumber = 0;
 193                 $this->mLastSection = '';
 194                 $this->mDTopen = false;
 195                 $this->mIncludeCount = array();
 196                 $this->mStripState = new StripState;
 197                 $this->mArgStack = false;
 198                 $this->mInPre = false;
 199                 $this->mInterwikiLinkHolders = array(
 200                         'texts' => array(),
 201                         'titles' => array()
 202                 );
 203                 $this->mLinkHolders = array(
 204                         'namespaces' => array(),
 205                         'dbkeys' => array(),
 206                         'queries' => array(),
 207                         'texts' => array(),
 208                         'titles' => array()
 209                 );
 210                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 211
 212                 /**
 213                  * Prefix for temporary replacement strings for the multipass parser.
 214                  * \x07 should never appear in input as it's disallowed in XML.
 215                  * Using it at the front also gives us a little extra robustness
 216                  * since it shouldn't match when butted up against identifier-like
 217                  * string constructs.
 218                  *
 219                  * Must not consist of all title characters, or else it will change
 220                  * the behaviour of <nowiki> in a link.
 221                  */
 222                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 223                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 224
 225                 # Clear these on every parse, bug 4549
 226                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 227
 228                 $this->mShowToc = true;
 229                 $this->mForceTocPosition = false;
 230                 $this->mIncludeSizes = array(
 231                         'post-expand' => 0,
 232                         'arg' => 0,
 233                 );
 234                 $this->mPPNodeCount = 0;
 235                 $this->mDefaultSort = false;
 236                 $this->mHeadings = array();
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         function setOutputType( $ot ) {
 243                 $this->mOutputType = $ot;
 244                 // Shortcut alias
 245                 $this->ot = array(
 246                         'html' => $ot == OT_HTML,
 247                         'wiki' => $ot == OT_WIKI,
 248                         'msg' => $ot == OT_MSG,
 249                         'pre' => $ot == OT_PREPROCESS,
 250                 );
 251         }
 252
 253         /**
 254          * Set the context title
 255          */
 256         function setTitle( $t ) {
 257                 if ( strval( $t->getFragment() ) !== '' ) {
 258                         # Strip the fragment to avoid various odd effects
 259                         $this->mTitle = clone $t;
 260                         $this->mTitle->setFragment( '' );
 261                 } else {
 262                         $this->mTitle = $t;
 263                 }
 264         }
 265
 266         /**
 267          * Accessor for mUniqPrefix.
 268          *
 269          * @public
 270          */
 271         function uniqPrefix() {
 272                 if( !isset( $this->mUniqPrefix ) ) {
 273                         // @fixme this is probably *horribly wrong*
 274                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 275                         // if this is called for a parser cache hit, the parser may not
 276                         // have ever been initialized in the first place.
 277                         // Not really sure what the heck is supposed to be going on here.
 278                         return '';
 279                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 280                 }
 281                 return $this->mUniqPrefix;
 282         }
 283
 284         /**
 285          * Convert wikitext to HTML
 286          * Do not call this function recursively.
 287          *
 288          * @param string $text Text we want to parse
 289          * @param Title &$title A title object
 290          * @param array $options
 291          * @param boolean $linestart
 292          * @param boolean $clearState
 293          * @param int $revid number to pass in {{REVISIONID}}
 294          * @return ParserOutput a ParserOutput
 295          */
 296         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 297                 /**
 298                  * First pass--just handle <nowiki> sections, pass the rest off
 299                  * to internalParse() which does all the real work.
 300                  */
 301
 302                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 303                 $fname = 'Parser::parse-' . wfGetCaller();
 304                 wfProfileIn( __METHOD__ );
 305                 wfProfileIn( $fname );
 306
 307                 if ( $clearState ) {
 308                         $this->clearState();
 309                 }
 310
 311                 $this->mOptions = $options;
 312                 $this->setTitle( $title );
 313                 $oldRevisionId = $this->mRevisionId;
 314                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 315                 if( $revid !== null ) {
 316                         $this->mRevisionId = $revid;
 317                         $this->mRevisionTimestamp = null;
 318                 }
 319                 $this->setOutputType( OT_HTML );
 320                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 321                 # No more strip!
 322                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 323                 $text = $this->internalParse( $text );
 324                 $text = $this->mStripState->unstripGeneral( $text );
 325
 326                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 327                 $fixtags = array(
 328                         # french spaces, last one Guillemet-left
 329                         # only if there is something before the space
 330                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 331                         # french spaces, Guillemet-right
 332                         '/(\\302\\253) /' => '\\1&nbsp;',
 333                 );
 334                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 335
 336                 # only once and last
 337                 $text = $this->doBlockLevels( $text, $linestart );
 338
 339                 $this->replaceLinkHolders( $text );
 340
 341                 # the position of the parserConvert() call should not be changed. it
 342                 # assumes that the links are all replaced and the only thing left
 343                 # is the <nowiki> mark.
 344                 # Side-effects: this calls $this->mOutput->setTitleText()
 345                 $text = $wgContLang->parserConvert( $text, $this );
 346
 347                 $text = $this->mStripState->unstripNoWiki( $text );
 348
 349                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 350
 351 //!JF Move to its own function
 352
 353                 $uniq_prefix = $this->mUniqPrefix;
 354                 $matches = array();
 355                 $elements = array_keys( $this->mTransparentTagHooks );
 356                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 357
 358                 foreach( $matches as $marker => $data ) {
 359                         list( $element, $content, $params, $tag ) = $data;
 360                         $tagName = strtolower( $element );
 361                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 362                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 363                                         array( $content, $params, $this ) );
 364                         } else {
 365                                 $output = $tag;
 366                         }
 367                         $this->mStripState->general->setPair( $marker, $output );
 368                 }
 369                 $text = $this->mStripState->unstripGeneral( $text );
 370
 371                 $text = Sanitizer::normalizeCharReferences( $text );
 372
 373                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 374                         $text = Parser::tidy($text);
 375                 } else {
 376                         # attempt to sanitize at least some nesting problems
 377                         # (bug #2702 and quite a few others)
 378                         $tidyregs = array(
 379                                 # ''Something [http://www.cool.com cool''] -->
 380                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 381                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 382                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 383                                 # fix up an anchor inside another anchor, only
 384                                 # at least for a single single nested link (bug 3695)
 385                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 386                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 387                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 388                                 # contains a div, so fix it up here; replace
 389                                 # div with escaped text
 390                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 391                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 392                                 # remove empty italic or bold tag pairs, some
 393                                 # introduced by rules above
 394                                 '/<([bi])><\/\\1>/' => '',
 395                         );
 396
 397                         $text = preg_replace(
 398                                 array_keys( $tidyregs ),
 399                                 array_values( $tidyregs ),
 400                                 $text );
 401                 }
 402
 403                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 404
 405                 # Information on include size limits, for the benefit of users who try to skirt them
 406                 if ( $this->mOptions->getEnableLimitReport() ) {
 407                         $max = $this->mOptions->getMaxIncludeSize();
 408                         $limitReport =
 409                                 "NewPP limit report\n" .
 410                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 411                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 412                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 413                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 414                         $text .= "\n<!-- \n$limitReport-->\n";
 415                 }
 416                 $this->mOutput->setText( $text );
 417                 $this->mRevisionId = $oldRevisionId;
 418                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 419                 wfProfileOut( $fname );
 420                 wfProfileOut( __METHOD__ );
 421
 422                 return $this->mOutput;
 423         }
 424
 425         /**
 426          * Recursive parser entry point that can be called from an extension tag
 427          * hook.
 428          */
 429         function recursiveTagParse( $text ) {
 430                 wfProfileIn( __METHOD__ );
 431                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 432                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 433                 $text = $this->internalParse( $text );
 434                 wfProfileOut( __METHOD__ );
 435                 return $text;
 436         }
 437
 438         /**
 439          * Expand templates and variables in the text, producing valid, static wikitext.
 440          * Also removes comments.
 441          */
 442         function preprocess( $text, $title, $options, $revid = null ) {
 443                 wfProfileIn( __METHOD__ );
 444                 $this->clearState();
 445                 $this->setOutputType( OT_PREPROCESS );
 446                 $this->mOptions = $options;
 447                 $this->setTitle( $title );
 448                 if( $revid !== null ) {
 449                         $this->mRevisionId = $revid;
 450                 }
 451                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 452                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 453                 $text = $this->replaceVariables( $text );
 454                 if ( $this->mOptions->getRemoveComments() ) {
 455                         $text = Sanitizer::removeHTMLcomments( $text );
 456                 }
 457                 $text = $this->mStripState->unstripBoth( $text );
 458                 wfProfileOut( __METHOD__ );
 459                 return $text;
 460         }
 461
 462         /**
 463          * Get a random string
 464          *
 465          * @private
 466          * @static
 467          */
 468         function getRandomString() {
 469                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 470         }
 471
 472         function &getTitle() { return $this->mTitle; }
 473         function getOptions() { return $this->mOptions; }
 474
 475         function getFunctionLang() {
 476                 global $wgLang, $wgContLang;
 477                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 478         }
 479
 480         /**
 481          * Replaces all occurrences of HTML-style comments and the given tags
 482          * in the text with a random marker and returns teh next text. The output
 483          * parameter $matches will be an associative array filled with data in
 484          * the form:
 485          *   'UNIQ-xxxxx' => array(
 486          *     'element',
 487          *     'tag content',
 488          *     array( 'param' => 'x' ),
 489          *     '<element param="x">tag content</element>' ) )
 490          *
 491          * @param $elements list of element names. Comments are always extracted.
 492          * @param $text Source text string.
 493          * @param $uniq_prefix
 494          *
 495          * @public
 496          * @static
 497          */
 498         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 499                 static $n = 1;
 500                 $stripped = '';
 501                 $matches = array();
 502
 503                 $taglist = implode( '|', $elements );
 504                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 505
 506                 while ( '' != $text ) {
 507                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 508                         $stripped .= $p[0];
 509                         if( count( $p ) < 5 ) {
 510                                 break;
 511                         }
 512                         if( count( $p ) > 5 ) {
 513                                 // comment
 514                                 $element    = $p[4];
 515                                 $attributes = '';
 516                                 $close      = '';
 517                                 $inside     = $p[5];
 518                         } else {
 519                                 // tag
 520                                 $element    = $p[1];
 521                                 $attributes = $p[2];
 522                                 $close      = $p[3];
 523                                 $inside     = $p[4];
 524                         }
 525
 526                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 527                         $stripped .= $marker;
 528
 529                         if ( $close === '/>' ) {
 530                                 // Empty element tag, <tag />
 531                                 $content = null;
 532                                 $text = $inside;
 533                                 $tail = null;
 534                         } else {
 535                                 if( $element == '!--' ) {
 536                                         $end = '/(-->)/';
 537                                 } else {
 538                                         $end = "/(<\\/$element\\s*>)/i";
 539                                 }
 540                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 541                                 $content = $q[0];
 542                                 if( count( $q ) < 3 ) {
 543                                         # No end tag -- let it run out to the end of the text.
 544                                         $tail = '';
 545                                         $text = '';
 546                                 } else {
 547                                         $tail = $q[1];
 548                                         $text = $q[2];
 549                                 }
 550                         }
 551
 552                         $matches[$marker] = array( $element,
 553                                 $content,
 554                                 Sanitizer::decodeTagAttributes( $attributes ),
 555                                 "<$element$attributes$close$content$tail" );
 556                 }
 557                 return $stripped;
 558         }
 559
 560         /**
 561          * Get a list of strippable XML-like elements
 562          */
 563         function getStripList() {
 564                 global $wgRawHtml;
 565                 $elements = $this->mStripList;
 566                 if( $wgRawHtml ) {
 567                         $elements[] = 'html';
 568                 }
 569                 if( $this->mOptions->getUseTeX() ) {
 570                         $elements[] = 'math';
 571                 }
 572                 return $elements;
 573         }
 574
 575         /**
 576          * @deprecated use replaceVariables
 577          */
 578         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 579                 return $text;
 580         }
 581
 582         /**
 583          * Restores pre, math, and other extensions removed by strip()
 584          *
 585          * always call unstripNoWiki() after this one
 586          * @private
 587          * @deprecated use $this->mStripState->unstrip()
 588          */
 589         function unstrip( $text, $state ) {
 590                 return $state->unstripGeneral( $text );
 591         }
 592
 593         /**
 594          * Always call this after unstrip() to preserve the order
 595          *
 596          * @private
 597          * @deprecated use $this->mStripState->unstrip()
 598          */
 599         function unstripNoWiki( $text, $state ) {
 600                 return $state->unstripNoWiki( $text );
 601         }
 602
 603         /**
 604          * @deprecated use $this->mStripState->unstripBoth()
 605          */
 606         function unstripForHTML( $text ) {
 607                 return $this->mStripState->unstripBoth( $text );
 608         }
 609
 610         /**
 611          * Add an item to the strip state
 612          * Returns the unique tag which must be inserted into the stripped text
 613          * The tag will be replaced with the original text in unstrip()
 614          *
 615          * @private
 616          */
 617         function insertStripItem( $text ) {
 618                 static $n = 0;
 619                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 620                 ++$n;
 621                 $this->mStripState->general->setPair( $rnd, $text );
 622                 return $rnd;
 623         }
 624
 625         /**
 626          * Interface with html tidy, used if $wgUseTidy = true.
 627          * If tidy isn't able to correct the markup, the original will be
 628          * returned in all its glory with a warning comment appended.
 629          *
 630          * Either the external tidy program or the in-process tidy extension
 631          * will be used depending on availability. Override the default
 632          * $wgTidyInternal setting to disable the internal if it's not working.
 633          *
 634          * @param string $text Hideous HTML input
 635          * @return string Corrected HTML output
 636          * @public
 637          * @static
 638          */
 639         function tidy( $text ) {
 640                 global $wgTidyInternal;
 641                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 642 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 643 '<head><title>test</title></head><body>'.$text.'</body></html>';
 644                 if( $wgTidyInternal ) {
 645                         $correctedtext = Parser::internalTidy( $wrappedtext );
 646                 } else {
 647                         $correctedtext = Parser::externalTidy( $wrappedtext );
 648                 }
 649                 if( is_null( $correctedtext ) ) {
 650                         wfDebug( "Tidy error detected!\n" );
 651                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 652                 }
 653                 return $correctedtext;
 654         }
 655
 656         /**
 657          * Spawn an external HTML tidy process and get corrected markup back from it.
 658          *
 659          * @private
 660          * @static
 661          */
 662         function externalTidy( $text ) {
 663                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 664                 $fname = 'Parser::externalTidy';
 665                 wfProfileIn( $fname );
 666
 667                 $cleansource = '';
 668                 $opts = ' -utf8';
 669
 670                 $descriptorspec = array(
 671                         0 => array('pipe', 'r'),
 672                         1 => array('pipe', 'w'),
 673                         2 => array('file', wfGetNull(), 'a')
 674                 );
 675                 $pipes = array();
 676                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 677                 if (is_resource($process)) {
 678                         // Theoretically, this style of communication could cause a deadlock
 679                         // here. If the stdout buffer fills up, then writes to stdin could
 680                         // block. This doesn't appear to happen with tidy, because tidy only
 681                         // writes to stdout after it's finished reading from stdin. Search
 682                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 683                         fwrite($pipes[0], $text);
 684                         fclose($pipes[0]);
 685                         while (!feof($pipes[1])) {
 686                                 $cleansource .= fgets($pipes[1], 1024);
 687                         }
 688                         fclose($pipes[1]);
 689                         proc_close($process);
 690                 }
 691
 692                 wfProfileOut( $fname );
 693
 694                 if( $cleansource == '' && $text != '') {
 695                         // Some kind of error happened, so we couldn't get the corrected text.
 696                         // Just give up; we'll use the source text and append a warning.
 697                         return null;
 698                 } else {
 699                         return $cleansource;
 700                 }
 701         }
 702
 703         /**
 704          * Use the HTML tidy PECL extension to use the tidy library in-process,
 705          * saving the overhead of spawning a new process.
 706          *
 707          * 'pear install tidy' should be able to compile the extension module.
 708          *
 709          * @private
 710          * @static
 711          */
 712         function internalTidy( $text ) {
 713                 global $wgTidyConf, $IP, $wgDebugTidy;
 714                 $fname = 'Parser::internalTidy';
 715                 wfProfileIn( $fname );
 716
 717                 $tidy = new tidy;
 718                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 719                 $tidy->cleanRepair();
 720                 if( $tidy->getStatus() == 2 ) {
 721                         // 2 is magic number for fatal error
 722                         // http://www.php.net/manual/en/function.tidy-get-status.php
 723                         $cleansource = null;
 724                 } else {
 725                         $cleansource = tidy_get_output( $tidy );
 726                 }
 727                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 728                         $cleansource .= "<!--\nTidy reports:\n" .
 729                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 730                                 "\n-->";
 731                 }
 732
 733                 wfProfileOut( $fname );
 734                 return $cleansource;
 735         }
 736
 737         /**
 738          * parse the wiki syntax used to render tables
 739          *
 740          * @private
 741          */
 742         function doTableStuff ( $text ) {
 743                 $fname = 'Parser::doTableStuff';
 744                 wfProfileIn( $fname );
 745
 746                 $lines = explode ( "\n" , $text );
 747                 $td_history = array (); // Is currently a td tag open?
 748                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 749                 $tr_history = array (); // Is currently a tr tag open?
 750                 $tr_attributes = array (); // history of tr attributes
 751                 $has_opened_tr = array(); // Did this table open a <tr> element?
 752                 $indent_level = 0; // indent level of the table
 753                 foreach ( $lines as $key => $line )
 754                 {
 755                         $line = trim ( $line );
 756
 757                         if( $line == '' ) { // empty line, go to next line
 758                                 continue;
 759                         }
 760                         $first_character = $line{0};
 761                         $matches = array();
 762
 763                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 764                                 // First check if we are starting a new table
 765                                 $indent_level = strlen( $matches[1] );
 766
 767                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 768                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 769
 770                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 771                                 array_push ( $td_history , false );
 772                                 array_push ( $last_tag_history , '' );
 773                                 array_push ( $tr_history , false );
 774                                 array_push ( $tr_attributes , '' );
 775                                 array_push ( $has_opened_tr , false );
 776                         } else if ( count ( $td_history ) == 0 ) {
 777                                 // Don't do any of the following
 778                                 continue;
 779                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 780                                 // We are ending a table
 781                                 $line = '</table>' . substr ( $line , 2 );
 782                                 $last_tag = array_pop ( $last_tag_history );
 783
 784                                 if ( !array_pop ( $has_opened_tr ) ) {
 785                                         $line = "<tr><td></td></tr>{$line}";
 786                                 }
 787
 788                                 if ( array_pop ( $tr_history ) ) {
 789                                         $line = "</tr>{$line}";
 790                                 }
 791
 792                                 if ( array_pop ( $td_history ) ) {
 793                                         $line = "</{$last_tag}>{$line}";
 794                                 }
 795                                 array_pop ( $tr_attributes );
 796                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 797                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 798                                 // Now we have a table row
 799                                 $line = preg_replace( '#^\|-+#', '', $line );
 800
 801                                 // Whats after the tag is now only attributes
 802                                 $attributes = $this->mStripState->unstripBoth( $line );
 803                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 804                                 array_pop ( $tr_attributes );
 805                                 array_push ( $tr_attributes , $attributes );
 806
 807                                 $line = '';
 808                                 $last_tag = array_pop ( $last_tag_history );
 809                                 array_pop ( $has_opened_tr );
 810                                 array_push ( $has_opened_tr , true );
 811
 812                                 if ( array_pop ( $tr_history ) ) {
 813                                         $line = '</tr>';
 814                                 }
 815
 816                                 if ( array_pop ( $td_history ) ) {
 817                                         $line = "</{$last_tag}>{$line}";
 818                                 }
 819
 820                                 $lines[$key] = $line;
 821                                 array_push ( $tr_history , false );
 822                                 array_push ( $td_history , false );
 823                                 array_push ( $last_tag_history , '' );
 824                         }
 825                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 826                                 // This might be cell elements, td, th or captions
 827                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 828                                         $first_character = '+';
 829                                         $line = substr ( $line , 1 );
 830                                 }
 831
 832                                 $line = substr ( $line , 1 );
 833
 834                                 if ( $first_character == '!' ) {
 835                                         $line = str_replace ( '!!' , '||' , $line );
 836                                 }
 837
 838                                 // Split up multiple cells on the same line.
 839                                 // FIXME : This can result in improper nesting of tags processed
 840                                 // by earlier parser steps, but should avoid splitting up eg
 841                                 // attribute values containing literal "||".
 842                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 843
 844                                 $lines[$key] = '';
 845
 846                                 // Loop through each table cell
 847                                 foreach ( $cells as $cell )
 848                                 {
 849                                         $previous = '';
 850                                         if ( $first_character != '+' )
 851                                         {
 852                                                 $tr_after = array_pop ( $tr_attributes );
 853                                                 if ( !array_pop ( $tr_history ) ) {
 854                                                         $previous = "<tr{$tr_after}>\n";
 855                                                 }
 856                                                 array_push ( $tr_history , true );
 857                                                 array_push ( $tr_attributes , '' );
 858                                                 array_pop ( $has_opened_tr );
 859                                                 array_push ( $has_opened_tr , true );
 860                                         }
 861
 862                                         $last_tag = array_pop ( $last_tag_history );
 863
 864                                         if ( array_pop ( $td_history ) ) {
 865                                                 $previous = "</{$last_tag}>{$previous}";
 866                                         }
 867
 868                                         if ( $first_character == '|' ) {
 869                                                 $last_tag = 'td';
 870                                         } else if ( $first_character == '!' ) {
 871                                                 $last_tag = 'th';
 872                                         } else if ( $first_character == '+' ) {
 873                                                 $last_tag = 'caption';
 874                                         } else {
 875                                                 $last_tag = '';
 876                                         }
 877
 878                                         array_push ( $last_tag_history , $last_tag );
 879
 880                                         // A cell could contain both parameters and data
 881                                         $cell_data = explode ( '|' , $cell , 2 );
 882
 883                                         // Bug 553: Note that a '|' inside an invalid link should not
 884                                         // be mistaken as delimiting cell parameters
 885                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 886                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 887                                         } else if ( count ( $cell_data ) == 1 )
 888                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 889                                         else {
 890                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 891                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 892                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 893                                         }
 894
 895                                         $lines[$key] .= $cell;
 896                                         array_push ( $td_history , true );
 897                                 }
 898                         }
 899                 }
 900
 901                 // Closing open td, tr && table
 902                 while ( count ( $td_history ) > 0 )
 903                 {
 904                         if ( array_pop ( $td_history ) ) {
 905                                 $lines[] = '</td>' ;
 906                         }
 907                         if ( array_pop ( $tr_history ) ) {
 908                                 $lines[] = '</tr>' ;
 909                         }
 910                         if ( !array_pop ( $has_opened_tr ) ) {
 911                                 $lines[] = "<tr><td></td></tr>" ;
 912                         }
 913
 914                         $lines[] = '</table>' ;
 915                 }
 916
 917                 $output = implode ( "\n" , $lines ) ;
 918
 919                 // special case: don't return empty table
 920                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 921                         $output = '';
 922                 }
 923
 924                 wfProfileOut( $fname );
 925
 926                 return $output;
 927         }
 928
 929         /**
 930          * Helper function for parse() that transforms wiki markup into
 931          * HTML. Only called for $mOutputType == OT_HTML.
 932          *
 933          * @private
 934          */
 935         function internalParse( $text ) {
 936                 $isMain = true;
 937                 $fname = 'Parser::internalParse';
 938                 wfProfileIn( $fname );
 939
 940                 # Hook to suspend the parser in this state
 941                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 942                         wfProfileOut( $fname );
 943                         return $text ;
 944                 }
 945
 946                 $text = $this->replaceVariables( $text );
 947                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 948                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 949
 950                 // Tables need to come after variable replacement for things to work
 951                 // properly; putting them before other transformations should keep
 952                 // exciting things like link expansions from showing up in surprising
 953                 // places.
 954                 $text = $this->doTableStuff( $text );
 955
 956                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 957
 958                 $text = $this->stripToc( $text );
 959                 $this->stripNoGallery( $text );
 960                 $text = $this->doHeadings( $text );
 961                 if($this->mOptions->getUseDynamicDates()) {
 962                         $df =& DateFormatter::getInstance();
 963                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 964                 }
 965                 $text = $this->doAllQuotes( $text );
 966                 $text = $this->replaceInternalLinks( $text );
 967                 $text = $this->replaceExternalLinks( $text );
 968
 969                 # replaceInternalLinks may sometimes leave behind
 970                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 971                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 972
 973                 $text = $this->doMagicLinks( $text );
 974                 $text = $this->formatHeadings( $text, $isMain );
 975
 976                 wfProfileOut( $fname );
 977                 return $text;
 978         }
 979
 980         /**
 981          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 982          * magic external links.
 983          *
 984          * @private
 985          */
 986         function doMagicLinks( $text ) {
 987                 wfProfileIn( __METHOD__ );
 988                 $text = preg_replace_callback(
 989                         '!(?:                           # Start cases
 990                             <a.*?</a> |                 # Skip link text
 991                             <.*?> |                     # Skip stuff inside HTML elements
 992                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 993                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 994                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 995                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 996                                      [0-9Xx]                 # check digit
 997                                    \b)
 998                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 999                 wfProfileOut( __METHOD__ );
1000                 return $text;
1001         }
1002
1003         function magicLinkCallback( $m ) {
1004                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1005                         # Skip HTML element
1006                         return $m[0];
1007                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1008                         $isbn = $m[2];
1009                         $num = strtr( $isbn, array(
1010                                 '-' => '',
1011                                 ' ' => '',
1012                                 'x' => 'X',
1013                         ));
1014                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1015                         $text = '<a href="' .
1016                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1017                                 "\" class=\"internal\">ISBN $isbn</a>";
1018                 } else {
1019                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1020                                 $keyword = 'RFC';
1021                                 $urlmsg = 'rfcurl';
1022                                 $id = $m[1];
1023                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1024                                 $keyword = 'PMID';
1025                                 $urlmsg = 'pubmedurl';
1026                                 $id = $m[1];
1027                         } else {
1028                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1029                                         substr($m[0], 0, 20 ) . '"' );
1030                         }
1031
1032                         $url = wfMsg( $urlmsg, $id);
1033                         $sk = $this->mOptions->getSkin();
1034                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1035                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1036                 }
1037                 return $text;
1038         }
1039
1040         /**
1041          * Parse headers and return html
1042          *
1043          * @private
1044          */
1045         function doHeadings( $text ) {
1046                 $fname = 'Parser::doHeadings';
1047                 wfProfileIn( $fname );
1048                 for ( $i = 6; $i >= 1; --$i ) {
1049                         $h = str_repeat( '=', $i );
1050                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1051                           "<h$i>\\1</h$i>", $text );
1052                 }
1053                 wfProfileOut( $fname );
1054                 return $text;
1055         }
1056
1057         /**
1058          * Replace single quotes with HTML markup
1059          * @private
1060          * @return string the altered text
1061          */
1062         function doAllQuotes( $text ) {
1063                 $fname = 'Parser::doAllQuotes';
1064                 wfProfileIn( $fname );
1065                 $outtext = '';
1066                 $lines = explode( "\n", $text );
1067                 foreach ( $lines as $line ) {
1068                         $outtext .= $this->doQuotes ( $line ) . "\n";
1069                 }
1070                 $outtext = substr($outtext, 0,-1);
1071                 wfProfileOut( $fname );
1072                 return $outtext;
1073         }
1074
1075         /**
1076          * Helper function for doAllQuotes()
1077          */
1078         public function doQuotes( $text ) {
1079                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1080                 if ( count( $arr ) == 1 )
1081                         return $text;
1082                 else
1083                 {
1084                         # First, do some preliminary work. This may shift some apostrophes from
1085                         # being mark-up to being text. It also counts the number of occurrences
1086                         # of bold and italics mark-ups.
1087                         $i = 0;
1088                         $numbold = 0;
1089                         $numitalics = 0;
1090                         foreach ( $arr as $r )
1091                         {
1092                                 if ( ( $i % 2 ) == 1 )
1093                                 {
1094                                         # If there are ever four apostrophes, assume the first is supposed to
1095                                         # be text, and the remaining three constitute mark-up for bold text.
1096                                         if ( strlen( $arr[$i] ) == 4 )
1097                                         {
1098                                                 $arr[$i-1] .= "'";
1099                                                 $arr[$i] = "'''";
1100                                         }
1101                                         # If there are more than 5 apostrophes in a row, assume they're all
1102                                         # text except for the last 5.
1103                                         else if ( strlen( $arr[$i] ) > 5 )
1104                                         {
1105                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1106                                                 $arr[$i] = "'''''";
1107                                         }
1108                                         # Count the number of occurrences of bold and italics mark-ups.
1109                                         # We are not counting sequences of five apostrophes.
1110                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1111                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1112                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1113                                 }
1114                                 $i++;
1115                         }
1116
1117                         # If there is an odd number of both bold and italics, it is likely
1118                         # that one of the bold ones was meant to be an apostrophe followed
1119                         # by italics. Which one we cannot know for certain, but it is more
1120                         # likely to be one that has a single-letter word before it.
1121                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1122                         {
1123                                 $i = 0;
1124                                 $firstsingleletterword = -1;
1125                                 $firstmultiletterword = -1;
1126                                 $firstspace = -1;
1127                                 foreach ( $arr as $r )
1128                                 {
1129                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1130                                         {
1131                                                 $x1 = substr ($arr[$i-1], -1);
1132                                                 $x2 = substr ($arr[$i-1], -2, 1);
1133                                                 if ($x1 == ' ') {
1134                                                         if ($firstspace == -1) $firstspace = $i;
1135                                                 } else if ($x2 == ' ') {
1136                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1137                                                 } else {
1138                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1139                                                 }
1140                                         }
1141                                         $i++;
1142                                 }
1143
1144                                 # If there is a single-letter word, use it!
1145                                 if ($firstsingleletterword > -1)
1146                                 {
1147                                         $arr [ $firstsingleletterword ] = "''";
1148                                         $arr [ $firstsingleletterword-1 ] .= "'";
1149                                 }
1150                                 # If not, but there's a multi-letter word, use that one.
1151                                 else if ($firstmultiletterword > -1)
1152                                 {
1153                                         $arr [ $firstmultiletterword ] = "''";
1154                                         $arr [ $firstmultiletterword-1 ] .= "'";
1155                                 }
1156                                 # ... otherwise use the first one that has neither.
1157                                 # (notice that it is possible for all three to be -1 if, for example,
1158                                 # there is only one pentuple-apostrophe in the line)
1159                                 else if ($firstspace > -1)
1160                                 {
1161                                         $arr [ $firstspace ] = "''";
1162                                         $arr [ $firstspace-1 ] .= "'";
1163                                 }
1164                         }
1165
1166                         # Now let's actually convert our apostrophic mush to HTML!
1167                         $output = '';
1168                         $buffer = '';
1169                         $state = '';
1170                         $i = 0;
1171                         foreach ($arr as $r)
1172                         {
1173                                 if (($i % 2) == 0)
1174                                 {
1175                                         if ($state == 'both')
1176                                                 $buffer .= $r;
1177                                         else
1178                                                 $output .= $r;
1179                                 }
1180                                 else
1181                                 {
1182                                         if (strlen ($r) == 2)
1183                                         {
1184                                                 if ($state == 'i')
1185                                                 { $output .= '</i>'; $state = ''; }
1186                                                 else if ($state == 'bi')
1187                                                 { $output .= '</i>'; $state = 'b'; }
1188                                                 else if ($state == 'ib')
1189                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1190                                                 else if ($state == 'both')
1191                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1192                                                 else # $state can be 'b' or ''
1193                                                 { $output .= '<i>'; $state .= 'i'; }
1194                                         }
1195                                         else if (strlen ($r) == 3)
1196                                         {
1197                                                 if ($state == 'b')
1198                                                 { $output .= '</b>'; $state = ''; }
1199                                                 else if ($state == 'bi')
1200                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1201                                                 else if ($state == 'ib')
1202                                                 { $output .= '</b>'; $state = 'i'; }
1203                                                 else if ($state == 'both')
1204                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1205                                                 else # $state can be 'i' or ''
1206                                                 { $output .= '<b>'; $state .= 'b'; }
1207                                         }
1208                                         else if (strlen ($r) == 5)
1209                                         {
1210                                                 if ($state == 'b')
1211                                                 { $output .= '</b><i>'; $state = 'i'; }
1212                                                 else if ($state == 'i')
1213                                                 { $output .= '</i><b>'; $state = 'b'; }
1214                                                 else if ($state == 'bi')
1215                                                 { $output .= '</i></b>'; $state = ''; }
1216                                                 else if ($state == 'ib')
1217                                                 { $output .= '</b></i>'; $state = ''; }
1218                                                 else if ($state == 'both')
1219                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1220                                                 else # ($state == '')
1221                                                 { $buffer = ''; $state = 'both'; }
1222                                         }
1223                                 }
1224                                 $i++;
1225                         }
1226                         # Now close all remaining tags.  Notice that the order is important.
1227                         if ($state == 'b' || $state == 'ib')
1228                                 $output .= '</b>';
1229                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1230                                 $output .= '</i>';
1231                         if ($state == 'bi')
1232                                 $output .= '</b>';
1233                         # There might be lonely ''''', so make sure we have a buffer
1234                         if ($state == 'both' && $buffer)
1235                                 $output .= '<b><i>'.$buffer.'</i></b>';
1236                         return $output;
1237                 }
1238         }
1239
1240         /**
1241          * Replace external links
1242          *
1243          * Note: this is all very hackish and the order of execution matters a lot.
1244          * Make sure to run maintenance/parserTests.php if you change this code.
1245          *
1246          * @private
1247          */
1248         function replaceExternalLinks( $text ) {
1249                 global $wgContLang;
1250                 $fname = 'Parser::replaceExternalLinks';
1251                 wfProfileIn( $fname );
1252
1253                 $sk = $this->mOptions->getSkin();
1254
1255                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1256
1257                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1258
1259                 $i = 0;
1260                 while ( $i<count( $bits ) ) {
1261                         $url = $bits[$i++];
1262                         $protocol = $bits[$i++];
1263                         $text = $bits[$i++];
1264                         $trail = $bits[$i++];
1265
1266                         # The characters '<' and '>' (which were escaped by
1267                         # removeHTMLtags()) should not be included in
1268                         # URLs, per RFC 2396.
1269                         $m2 = array();
1270                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1271                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1272                                 $url = substr($url, 0, $m2[0][1]);
1273                         }
1274
1275                         # If the link text is an image URL, replace it with an <img> tag
1276                         # This happened by accident in the original parser, but some people used it extensively
1277                         $img = $this->maybeMakeExternalImage( $text );
1278                         if ( $img !== false ) {
1279                                 $text = $img;
1280                         }
1281
1282                         $dtrail = '';
1283
1284                         # Set linktype for CSS - if URL==text, link is essentially free
1285                         $linktype = ($text == $url) ? 'free' : 'text';
1286
1287                         # No link text, e.g. [http://domain.tld/some.link]
1288                         if ( $text == '' ) {
1289                                 # Autonumber if allowed. See bug #5918
1290                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1291                                         $text = '[' . ++$this->mAutonumber . ']';
1292                                         $linktype = 'autonumber';
1293                                 } else {
1294                                         # Otherwise just use the URL
1295                                         $text = htmlspecialchars( $url );
1296                                         $linktype = 'free';
1297                                 }
1298                         } else {
1299                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1300                                 # Check for trail
1301                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1302                         }
1303
1304                         $text = $wgContLang->markNoConversion($text);
1305
1306                         $url = Sanitizer::cleanUrl( $url );
1307
1308                         # Process the trail (i.e. everything after this link up until start of the next link),
1309                         # replacing any non-bracketed links
1310                         $trail = $this->replaceFreeExternalLinks( $trail );
1311
1312                         # Use the encoded URL
1313                         # This means that users can paste URLs directly into the text
1314                         # Funny characters like &ouml; aren't valid in URLs anyway
1315                         # This was changed in August 2004
1316                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1317
1318                         # Register link in the output object.
1319                         # Replace unnecessary URL escape codes with the referenced character
1320                         # This prevents spammers from hiding links from the filters
1321                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1322                         $this->mOutput->addExternalLink( $pasteurized );
1323                 }
1324
1325                 wfProfileOut( $fname );
1326                 return $s;
1327         }
1328
1329         /**
1330          * Replace anything that looks like a URL with a link
1331          * @private
1332          */
1333         function replaceFreeExternalLinks( $text ) {
1334                 global $wgContLang;
1335                 $fname = 'Parser::replaceFreeExternalLinks';
1336                 wfProfileIn( $fname );
1337
1338                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1339                 $s = array_shift( $bits );
1340                 $i = 0;
1341
1342                 $sk = $this->mOptions->getSkin();
1343
1344                 while ( $i < count( $bits ) ){
1345                         $protocol = $bits[$i++];
1346                         $remainder = $bits[$i++];
1347
1348                         $m = array();
1349                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1350                                 # Found some characters after the protocol that look promising
1351                                 $url = $protocol . $m[1];
1352                                 $trail = $m[2];
1353
1354                                 # special case: handle urls as url args:
1355                                 # http://www.example.com/foo?=http://www.example.com/bar
1356                                 if(strlen($trail) == 0 &&
1357                                         isset($bits[$i]) &&
1358                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1359                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1360                                 {
1361                                         # add protocol, arg
1362                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1363                                         $i += 2;
1364                                         $trail = $m[2];
1365                                 }
1366
1367                                 # The characters '<' and '>' (which were escaped by
1368                                 # removeHTMLtags()) should not be included in
1369                                 # URLs, per RFC 2396.
1370                                 $m2 = array();
1371                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1372                                         $trail = substr($url, $m2[0][1]) . $trail;
1373                                         $url = substr($url, 0, $m2[0][1]);
1374                                 }
1375
1376                                 # Move trailing punctuation to $trail
1377                                 $sep = ',;\.:!?';
1378                                 # If there is no left bracket, then consider right brackets fair game too
1379                                 if ( strpos( $url, '(' ) === false ) {
1380                                         $sep .= ')';
1381                                 }
1382
1383                                 $numSepChars = strspn( strrev( $url ), $sep );
1384                                 if ( $numSepChars ) {
1385                                         $trail = substr( $url, -$numSepChars ) . $trail;
1386                                         $url = substr( $url, 0, -$numSepChars );
1387                                 }
1388
1389                                 $url = Sanitizer::cleanUrl( $url );
1390
1391                                 # Is this an external image?
1392                                 $text = $this->maybeMakeExternalImage( $url );
1393                                 if ( $text === false ) {
1394                                         # Not an image, make a link
1395                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1396                                         # Register it in the output object...
1397                                         # Replace unnecessary URL escape codes with their equivalent characters
1398                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1399                                         $this->mOutput->addExternalLink( $pasteurized );
1400                                 }
1401                                 $s .= $text . $trail;
1402                         } else {
1403                                 $s .= $protocol . $remainder;
1404                         }
1405                 }
1406                 wfProfileOut( $fname );
1407                 return $s;
1408         }
1409
1410         /**
1411          * Replace unusual URL escape codes with their equivalent characters
1412          * @param string
1413          * @return string
1414          * @static
1415          * @todo  This can merge genuinely required bits in the path or query string,
1416          *        breaking legit URLs. A proper fix would treat the various parts of
1417          *        the URL differently; as a workaround, just use the output for
1418          *        statistical records, not for actual linking/output.
1419          */
1420         static function replaceUnusualEscapes( $url ) {
1421                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1422                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1423         }
1424
1425         /**
1426          * Callback function used in replaceUnusualEscapes().
1427          * Replaces unusual URL escape codes with their equivalent character
1428          * @static
1429          * @private
1430          */
1431         private static function replaceUnusualEscapesCallback( $matches ) {
1432                 $char = urldecode( $matches[0] );
1433                 $ord = ord( $char );
1434                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1435                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1436                         // No, shouldn't be escaped
1437                         return $char;
1438                 } else {
1439                         // Yes, leave it escaped
1440                         return $matches[0];
1441                 }
1442         }
1443
1444         /**
1445          * make an image if it's allowed, either through the global
1446          * option or through the exception
1447          * @private
1448          */
1449         function maybeMakeExternalImage( $url ) {
1450                 $sk = $this->mOptions->getSkin();
1451                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1452                 $imagesexception = !empty($imagesfrom);
1453                 $text = false;
1454                 if ( $this->mOptions->getAllowExternalImages()
1455                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1456                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1457                                 # Image found
1458                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1459                         }
1460                 }
1461                 return $text;
1462         }
1463
1464         /**
1465          * Process [[ ]] wikilinks
1466          *
1467          * @private
1468          */
1469         function replaceInternalLinks( $s ) {
1470                 global $wgContLang;
1471                 static $fname = 'Parser::replaceInternalLinks' ;
1472
1473                 wfProfileIn( $fname );
1474
1475                 wfProfileIn( $fname.'-setup' );
1476                 static $tc = FALSE;
1477                 # the % is needed to support urlencoded titles as well
1478                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1479
1480                 $sk = $this->mOptions->getSkin();
1481
1482                 #split the entire text string on occurences of [[
1483                 $a = explode( '[[', ' ' . $s );
1484                 #get the first element (all text up to first [[), and remove the space we added
1485                 $s = array_shift( $a );
1486                 $s = substr( $s, 1 );
1487
1488                 # Match a link having the form [[namespace:link|alternate]]trail
1489                 static $e1 = FALSE;
1490                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1491                 # Match cases where there is no "]]", which might still be images
1492                 static $e1_img = FALSE;
1493                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1494
1495                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1496                 $e2 = null;
1497                 if ( $useLinkPrefixExtension ) {
1498                         # Match the end of a line for a word that's not followed by whitespace,
1499                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1500                         $e2 = wfMsgForContent( 'linkprefix' );
1501                 }
1502
1503                 if( is_null( $this->mTitle ) ) {
1504                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1505                 }
1506                 $nottalk = !$this->mTitle->isTalkPage();
1507
1508                 if ( $useLinkPrefixExtension ) {
1509                         $m = array();
1510                         if ( preg_match( $e2, $s, $m ) ) {
1511                                 $first_prefix = $m[2];
1512                         } else {
1513                                 $first_prefix = false;
1514                         }
1515                 } else {
1516                         $prefix = '';
1517                 }
1518
1519                 if($wgContLang->hasVariants()) {
1520                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1521                 } else {
1522                         $selflink = array($this->mTitle->getPrefixedText());
1523                 }
1524                 $useSubpages = $this->areSubpagesAllowed();
1525                 wfProfileOut( $fname.'-setup' );
1526
1527                 # Loop for each link
1528                 for ($k = 0; isset( $a[$k] ); $k++) {
1529                         $line = $a[$k];
1530                         if ( $useLinkPrefixExtension ) {
1531                                 wfProfileIn( $fname.'-prefixhandling' );
1532                                 if ( preg_match( $e2, $s, $m ) ) {
1533                                         $prefix = $m[2];
1534                                         $s = $m[1];
1535                                 } else {
1536                                         $prefix='';
1537                                 }
1538                                 # first link
1539                                 if($first_prefix) {
1540                                         $prefix = $first_prefix;
1541                                         $first_prefix = false;
1542                                 }
1543                                 wfProfileOut( $fname.'-prefixhandling' );
1544                         }
1545
1546                         $might_be_img = false;
1547
1548                         wfProfileIn( "$fname-e1" );
1549                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1550                                 $text = $m[2];
1551                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1552                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1553                                 # the real problem is with the $e1 regex
1554                                 # See bug 1300.
1555                                 #
1556                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1557                                 # and no image is in sight. See bug 2095.
1558                                 #
1559                                 if( $text !== '' &&
1560                                         substr( $m[3], 0, 1 ) === ']' &&
1561                                         strpos($text, '[') !== false
1562                                 )
1563                                 {
1564                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1565                                         $m[3] = substr( $m[3], 1 );
1566                                 }
1567                                 # fix up urlencoded title texts
1568                                 if( strpos( $m[1], '%' ) !== false ) {
1569                                         # Should anchors '#' also be rejected?
1570                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1571                                 }
1572                                 $trail = $m[3];
1573                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1574                                 $might_be_img = true;
1575                                 $text = $m[2];
1576                                 if ( strpos( $m[1], '%' ) !== false ) {
1577                                         $m[1] = urldecode($m[1]);
1578                                 }
1579                                 $trail = "";
1580                         } else { # Invalid form; output directly
1581                                 $s .= $prefix . '[[' . $line ;
1582                                 wfProfileOut( "$fname-e1" );
1583                                 continue;
1584                         }
1585                         wfProfileOut( "$fname-e1" );
1586                         wfProfileIn( "$fname-misc" );
1587
1588                         # Don't allow internal links to pages containing
1589                         # PROTO: where PROTO is a valid URL protocol; these
1590                         # should be external links.
1591                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1592                                 $s .= $prefix . '[[' . $line ;
1593                                 continue;
1594                         }
1595
1596                         # Make subpage if necessary
1597                         if( $useSubpages ) {
1598                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1599                         } else {
1600                                 $link = $m[1];
1601                         }
1602
1603                         $noforce = (substr($m[1], 0, 1) != ':');
1604                         if (!$noforce) {
1605                                 # Strip off leading ':'
1606                                 $link = substr($link, 1);
1607                         }
1608
1609                         wfProfileOut( "$fname-misc" );
1610                         wfProfileIn( "$fname-title" );
1611                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1612                         if( !$nt ) {
1613                                 $s .= $prefix . '[[' . $line;
1614                                 wfProfileOut( "$fname-title" );
1615                                 continue;
1616                         }
1617
1618                         $ns = $nt->getNamespace();
1619                         $iw = $nt->getInterWiki();
1620                         wfProfileOut( "$fname-title" );
1621
1622                         if ($might_be_img) { # if this is actually an invalid link
1623                                 wfProfileIn( "$fname-might_be_img" );
1624                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1625                                         $found = false;
1626                                         while (isset ($a[$k+1]) ) {
1627                                                 #look at the next 'line' to see if we can close it there
1628                                                 $spliced = array_splice( $a, $k + 1, 1 );
1629                                                 $next_line = array_shift( $spliced );
1630                                                 $m = explode( ']]', $next_line, 3 );
1631                                                 if ( count( $m ) == 3 ) {
1632                                                         # the first ]] closes the inner link, the second the image
1633                                                         $found = true;
1634                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1635                                                         $trail = $m[2];
1636                                                         break;
1637                                                 } elseif ( count( $m ) == 2 ) {
1638                                                         #if there's exactly one ]] that's fine, we'll keep looking
1639                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1640                                                 } else {
1641                                                         #if $next_line is invalid too, we need look no further
1642                                                         $text .= '[[' . $next_line;
1643                                                         break;
1644                                                 }
1645                                         }
1646                                         if ( !$found ) {
1647                                                 # we couldn't find the end of this imageLink, so output it raw
1648                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1649                                                 $text = $this->replaceInternalLinks($text);
1650                                                 $s .= "{$prefix}[[$link|$text";
1651                                                 # note: no $trail, because without an end, there *is* no trail
1652                                                 wfProfileOut( "$fname-might_be_img" );
1653                                                 continue;
1654                                         }
1655                                 } else { #it's not an image, so output it raw
1656                                         $s .= "{$prefix}[[$link|$text";
1657                                         # note: no $trail, because without an end, there *is* no trail
1658                                         wfProfileOut( "$fname-might_be_img" );
1659                                         continue;
1660                                 }
1661                                 wfProfileOut( "$fname-might_be_img" );
1662                         }
1663
1664                         $wasblank = ( '' == $text );
1665                         if( $wasblank ) $text = $link;
1666
1667                         # Link not escaped by : , create the various objects
1668                         if( $noforce ) {
1669
1670                                 # Interwikis
1671                                 wfProfileIn( "$fname-interwiki" );
1672                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1673                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1674                                         $s = rtrim($s . $prefix);
1675                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1676                                         wfProfileOut( "$fname-interwiki" );
1677                                         continue;
1678                                 }
1679                                 wfProfileOut( "$fname-interwiki" );
1680
1681                                 if ( $ns == NS_IMAGE ) {
1682                                         wfProfileIn( "$fname-image" );
1683                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1684                                                 # recursively parse links inside the image caption
1685                                                 # actually, this will parse them in any other parameters, too,
1686                                                 # but it might be hard to fix that, and it doesn't matter ATM
1687                                                 $text = $this->replaceExternalLinks($text);
1688                                                 $text = $this->replaceInternalLinks($text);
1689
1690                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1691                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1692                                                 $this->mOutput->addImage( $nt->getDBkey() );
1693
1694                                                 wfProfileOut( "$fname-image" );
1695                                                 continue;
1696                                         } else {
1697                                                 # We still need to record the image's presence on the page
1698                                                 $this->mOutput->addImage( $nt->getDBkey() );
1699                                         }
1700                                         wfProfileOut( "$fname-image" );
1701
1702                                 }
1703
1704                                 if ( $ns == NS_CATEGORY ) {
1705                                         wfProfileIn( "$fname-category" );
1706                                         $s = rtrim($s . "\n"); # bug 87
1707
1708                                         if ( $wasblank ) {
1709                                                 $sortkey = $this->getDefaultSort();
1710                                         } else {
1711                                                 $sortkey = $text;
1712                                         }
1713                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1714                                         $sortkey = str_replace( "\n", '', $sortkey );
1715                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1716                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1717
1718                                         /**
1719                                          * Strip the whitespace Category links produce, see bug 87
1720                                          * @todo We might want to use trim($tmp, "\n") here.
1721                                          */
1722                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1723
1724                                         wfProfileOut( "$fname-category" );
1725                                         continue;
1726                                 }
1727                         }
1728
1729                         # Self-link checking
1730                         if( $nt->getFragment() === '' ) {
1731                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1732                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1733                                         continue;
1734                                 }
1735                         }
1736
1737                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1738                         if( $ns == NS_MEDIA ) {
1739                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1740                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1741                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1742                                 $this->mOutput->addImage( $nt->getDBkey() );
1743                                 continue;
1744                         } elseif( $ns == NS_SPECIAL ) {
1745                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1746                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1747                                 } else {
1748                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1749                                 }
1750                                 continue;
1751                         } elseif( $ns == NS_IMAGE ) {
1752                                 $img = wfFindFile( $nt );
1753                                 if( $img ) {
1754                                         // Force a blue link if the file exists; may be a remote
1755                                         // upload on the shared repository, and we want to see its
1756                                         // auto-generated page.
1757                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1758                                         $this->mOutput->addLink( $nt );
1759                                         continue;
1760                                 }
1761                         }
1762                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1763                 }
1764                 wfProfileOut( $fname );
1765                 return $s;
1766         }
1767
1768         /**
1769          * Make a link placeholder. The text returned can be later resolved to a real link with
1770          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1771          * parsing of interwiki links, and secondly to allow all existence checks and
1772          * article length checks (for stub links) to be bundled into a single query.
1773          *
1774          */
1775         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1776                 wfProfileIn( __METHOD__ );
1777                 if ( ! is_object($nt) ) {
1778                         # Fail gracefully
1779                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1780                 } else {
1781                         # Separate the link trail from the rest of the link
1782                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1783
1784                         if ( $nt->isExternal() ) {
1785                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1786                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1787                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1788                         } else {
1789                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1790                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1791                                 $this->mLinkHolders['queries'][] = $query;
1792                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1793                                 $this->mLinkHolders['titles'][] = $nt;
1794
1795                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1796                         }
1797                 }
1798                 wfProfileOut( __METHOD__ );
1799                 return $retVal;
1800         }
1801
1802         /**
1803          * Render a forced-blue link inline; protect against double expansion of
1804          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1805          * Since this little disaster has to split off the trail text to avoid
1806          * breaking URLs in the following text without breaking trails on the
1807          * wiki links, it's been made into a horrible function.
1808          *
1809          * @param Title $nt
1810          * @param string $text
1811          * @param string $query
1812          * @param string $trail
1813          * @param string $prefix
1814          * @return string HTML-wikitext mix oh yuck
1815          */
1816         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1817                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1818                 $sk = $this->mOptions->getSkin();
1819                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1820                 return $this->armorLinks( $link ) . $trail;
1821         }
1822
1823         /**
1824          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1825          * going to go through further parsing steps before inline URL expansion.
1826          *
1827          * In particular this is important when using action=render, which causes
1828          * full URLs to be included.
1829          *
1830          * Oh man I hate our multi-layer parser!
1831          *
1832          * @param string more-or-less HTML
1833          * @return string less-or-more HTML with NOPARSE bits
1834          */
1835         function armorLinks( $text ) {
1836                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1837                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1838         }
1839
1840         /**
1841          * Return true if subpage links should be expanded on this page.
1842          * @return bool
1843          */
1844         function areSubpagesAllowed() {
1845                 # Some namespaces don't allow subpages
1846                 global $wgNamespacesWithSubpages;
1847                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1848         }
1849
1850         /**
1851          * Handle link to subpage if necessary
1852          * @param string $target the source of the link
1853          * @param string &$text the link text, modified as necessary
1854          * @return string the full name of the link
1855          * @private
1856          */
1857         function maybeDoSubpageLink($target, &$text) {
1858                 # Valid link forms:
1859                 # Foobar -- normal
1860                 # :Foobar -- override special treatment of prefix (images, language links)
1861                 # /Foobar -- convert to CurrentPage/Foobar
1862                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1863                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1864                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1865
1866                 $fname = 'Parser::maybeDoSubpageLink';
1867                 wfProfileIn( $fname );
1868                 $ret = $target; # default return value is no change
1869
1870                 # Some namespaces don't allow subpages,
1871                 # so only perform processing if subpages are allowed
1872                 if( $this->areSubpagesAllowed() ) {
1873                         $hash = strpos( $target, '#' );
1874                         if( $hash !== false ) {
1875                                 $suffix = substr( $target, $hash );
1876                                 $target = substr( $target, 0, $hash );
1877                         } else {
1878                                 $suffix = '';
1879                         }
1880                         # bug 7425
1881                         $target = trim( $target );
1882                         # Look at the first character
1883                         if( $target != '' && $target{0} == '/' ) {
1884                                 # / at end means we don't want the slash to be shown
1885                                 $m = array();
1886                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1887                                 if( $trailingSlashes ) {
1888                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1889                                 } else {
1890                                         $noslash = substr( $target, 1 );
1891                                 }
1892
1893                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1894                                 if( '' === $text ) {
1895                                         $text = $target . $suffix;
1896                                 } # this might be changed for ugliness reasons
1897                         } else {
1898                                 # check for .. subpage backlinks
1899                                 $dotdotcount = 0;
1900                                 $nodotdot = $target;
1901                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1902                                         ++$dotdotcount;
1903                                         $nodotdot = substr( $nodotdot, 3 );
1904                                 }
1905                                 if($dotdotcount > 0) {
1906                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1907                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1908                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1909                                                 # / at the end means don't show full path
1910                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1911                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1912                                                         if( '' === $text ) {
1913                                                                 $text = $nodotdot . $suffix;
1914                                                         }
1915                                                 }
1916                                                 $nodotdot = trim( $nodotdot );
1917                                                 if( $nodotdot != '' ) {
1918                                                         $ret .= '/' . $nodotdot;
1919                                                 }
1920                                                 $ret .= $suffix;
1921                                         }
1922                                 }
1923                         }
1924                 }
1925
1926                 wfProfileOut( $fname );
1927                 return $ret;
1928         }
1929
1930         /**#@+
1931          * Used by doBlockLevels()
1932          * @private
1933          */
1934         /* private */ function closeParagraph() {
1935                 $result = '';
1936                 if ( '' != $this->mLastSection ) {
1937                         $result = '</' . $this->mLastSection  . ">\n";
1938                 }
1939                 $this->mInPre = false;
1940                 $this->mLastSection = '';
1941                 return $result;
1942         }
1943         # getCommon() returns the length of the longest common substring
1944         # of both arguments, starting at the beginning of both.
1945         #
1946         /* private */ function getCommon( $st1, $st2 ) {
1947                 $fl = strlen( $st1 );
1948                 $shorter = strlen( $st2 );
1949                 if ( $fl < $shorter ) { $shorter = $fl; }
1950
1951                 for ( $i = 0; $i < $shorter; ++$i ) {
1952                         if ( $st1{$i} != $st2{$i} ) { break; }
1953                 }
1954                 return $i;
1955         }
1956         # These next three functions open, continue, and close the list
1957         # element appropriate to the prefix character passed into them.
1958         #
1959         /* private */ function openList( $char ) {
1960                 $result = $this->closeParagraph();
1961
1962                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1963                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1964                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1965                 else if ( ';' == $char ) {
1966                         $result .= '<dl><dt>';
1967                         $this->mDTopen = true;
1968                 }
1969                 else { $result = '<!-- ERR 1 -->'; }
1970
1971                 return $result;
1972         }
1973
1974         /* private */ function nextItem( $char ) {
1975                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1976                 else if ( ':' == $char || ';' == $char ) {
1977                         $close = '</dd>';
1978                         if ( $this->mDTopen ) { $close = '</dt>'; }
1979                         if ( ';' == $char ) {
1980                                 $this->mDTopen = true;
1981                                 return $close . '<dt>';
1982                         } else {
1983                                 $this->mDTopen = false;
1984                                 return $close . '<dd>';
1985                         }
1986                 }
1987                 return '<!-- ERR 2 -->';
1988         }
1989
1990         /* private */ function closeList( $char ) {
1991                 if ( '*' == $char ) { $text = '</li></ul>'; }
1992                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1993                 else if ( ':' == $char ) {
1994                         if ( $this->mDTopen ) {
1995                                 $this->mDTopen = false;
1996                                 $text = '</dt></dl>';
1997                         } else {
1998                                 $text = '</dd></dl>';
1999                         }
2000                 }
2001                 else {  return '<!-- ERR 3 -->'; }
2002                 return $text."\n";
2003         }
2004         /**#@-*/
2005
2006         /**
2007          * Make lists from lines starting with ':', '*', '#', etc.
2008          *
2009          * @private
2010          * @return string the lists rendered as HTML
2011          */
2012         function doBlockLevels( $text, $linestart ) {
2013                 $fname = 'Parser::doBlockLevels';
2014                 wfProfileIn( $fname );
2015
2016                 # Parsing through the text line by line.  The main thing
2017                 # happening here is handling of block-level elements p, pre,
2018                 # and making lists from lines starting with * # : etc.
2019                 #
2020                 $textLines = explode( "\n", $text );
2021
2022                 $lastPrefix = $output = '';
2023                 $this->mDTopen = $inBlockElem = false;
2024                 $prefixLength = 0;
2025                 $paragraphStack = false;
2026
2027                 if ( !$linestart ) {
2028                         $output .= array_shift( $textLines );
2029                 }
2030                 foreach ( $textLines as $oLine ) {
2031                         $lastPrefixLength = strlen( $lastPrefix );
2032                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2033                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2034                         if ( !$this->mInPre ) {
2035                                 # Multiple prefixes may abut each other for nested lists.
2036                                 $prefixLength = strspn( $oLine, '*#:;' );
2037                                 $pref = substr( $oLine, 0, $prefixLength );
2038
2039                                 # eh?
2040                                 $pref2 = str_replace( ';', ':', $pref );
2041                                 $t = substr( $oLine, $prefixLength );
2042                                 $this->mInPre = !empty($preOpenMatch);
2043                         } else {
2044                                 # Don't interpret any other prefixes in preformatted text
2045                                 $prefixLength = 0;
2046                                 $pref = $pref2 = '';
2047                                 $t = $oLine;
2048                         }
2049
2050                         # List generation
2051                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2052                                 # Same as the last item, so no need to deal with nesting or opening stuff
2053                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2054                                 $paragraphStack = false;
2055
2056                                 if ( substr( $pref, -1 ) == ';') {
2057                                         # The one nasty exception: definition lists work like this:
2058                                         # ; title : definition text
2059                                         # So we check for : in the remainder text to split up the
2060                                         # title and definition, without b0rking links.
2061                                         $term = $t2 = '';
2062                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2063                                                 $t = $t2;
2064                                                 $output .= $term . $this->nextItem( ':' );
2065                                         }
2066                                 }
2067                         } elseif( $prefixLength || $lastPrefixLength ) {
2068                                 # Either open or close a level...
2069                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2070                                 $paragraphStack = false;
2071
2072                                 while( $commonPrefixLength < $lastPrefixLength ) {
2073                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2074                                         --$lastPrefixLength;
2075                                 }
2076                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2077                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2078                                 }
2079                                 while ( $prefixLength > $commonPrefixLength ) {
2080                                         $char = substr( $pref, $commonPrefixLength, 1 );
2081                                         $output .= $this->openList( $char );
2082
2083                                         if ( ';' == $char ) {
2084                                                 # FIXME: This is dupe of code above
2085                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2086                                                         $t = $t2;
2087                                                         $output .= $term . $this->nextItem( ':' );
2088                                                 }
2089                                         }
2090                                         ++$commonPrefixLength;
2091                                 }
2092                                 $lastPrefix = $pref2;
2093                         }
2094                         if( 0 == $prefixLength ) {
2095                                 wfProfileIn( "$fname-paragraph" );
2096                                 # No prefix (not in list)--go to paragraph mode
2097                                 // XXX: use a stack for nestable elements like span, table and div
2098                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2099                                 $closematch = preg_match(
2100                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2101                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2102                                 if ( $openmatch or $closematch ) {
2103                                         $paragraphStack = false;
2104                                         # TODO bug 5718: paragraph closed
2105                                         $output .= $this->closeParagraph();
2106                                         if ( $preOpenMatch and !$preCloseMatch ) {
2107                                                 $this->mInPre = true;
2108                                         }
2109                                         if ( $closematch ) {
2110                                                 $inBlockElem = false;
2111                                         } else {
2112                                                 $inBlockElem = true;
2113                                         }
2114                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2115                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2116                                                 // pre
2117                                                 if ($this->mLastSection != 'pre') {
2118                                                         $paragraphStack = false;
2119                                                         $output .= $this->closeParagraph().'<pre>';
2120                                                         $this->mLastSection = 'pre';
2121                                                 }
2122                                                 $t = substr( $t, 1 );
2123                                         } else {
2124                                                 // paragraph
2125                                                 if ( '' == trim($t) ) {
2126                                                         if ( $paragraphStack ) {
2127                                                                 $output .= $paragraphStack.'<br />';
2128                                                                 $paragraphStack = false;
2129                                                                 $this->mLastSection = 'p';
2130                                                         } else {
2131                                                                 if ($this->mLastSection != 'p' ) {
2132                                                                         $output .= $this->closeParagraph();
2133                                                                         $this->mLastSection = '';
2134                                                                         $paragraphStack = '<p>';
2135                                                                 } else {
2136                                                                         $paragraphStack = '</p><p>';
2137                                                                 }
2138                                                         }
2139                                                 } else {
2140                                                         if ( $paragraphStack ) {
2141                                                                 $output .= $paragraphStack;
2142                                                                 $paragraphStack = false;
2143                                                                 $this->mLastSection = 'p';
2144                                                         } else if ($this->mLastSection != 'p') {
2145                                                                 $output .= $this->closeParagraph().'<p>';
2146                                                                 $this->mLastSection = 'p';
2147                                                         }
2148                                                 }
2149                                         }
2150                                 }
2151                                 wfProfileOut( "$fname-paragraph" );
2152                         }
2153                         // somewhere above we forget to get out of pre block (bug 785)
2154                         if($preCloseMatch && $this->mInPre) {
2155                                 $this->mInPre = false;
2156                         }
2157                         if ($paragraphStack === false) {
2158                                 $output .= $t."\n";
2159                         }
2160                 }
2161                 while ( $prefixLength ) {
2162                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2163                         --$prefixLength;
2164                 }
2165                 if ( '' != $this->mLastSection ) {
2166                         $output .= '</' . $this->mLastSection . '>';
2167                         $this->mLastSection = '';
2168                 }
2169
2170                 wfProfileOut( $fname );
2171                 return $output;
2172         }
2173
2174         /**
2175          * Split up a string on ':', ignoring any occurences inside tags
2176          * to prevent illegal overlapping.
2177          * @param string $str the string to split
2178          * @param string &$before set to everything before the ':'
2179          * @param string &$after set to everything after the ':'
2180          * return string the position of the ':', or false if none found
2181          */
2182         function findColonNoLinks($str, &$before, &$after) {
2183                 $fname = 'Parser::findColonNoLinks';
2184                 wfProfileIn( $fname );
2185
2186                 $pos = strpos( $str, ':' );
2187                 if( $pos === false ) {
2188                         // Nothing to find!
2189                         wfProfileOut( $fname );
2190                         return false;
2191                 }
2192
2193                 $lt = strpos( $str, '<' );
2194                 if( $lt === false || $lt > $pos ) {
2195                         // Easy; no tag nesting to worry about
2196                         $before = substr( $str, 0, $pos );
2197                         $after = substr( $str, $pos+1 );
2198                         wfProfileOut( $fname );
2199                         return $pos;
2200                 }
2201
2202                 // Ugly state machine to walk through avoiding tags.
2203                 $state = self::COLON_STATE_TEXT;
2204                 $stack = 0;
2205                 $len = strlen( $str );
2206                 for( $i = 0; $i < $len; $i++ ) {
2207                         $c = $str{$i};
2208
2209                         switch( $state ) {
2210                         // (Using the number is a performance hack for common cases)
2211                         case 0: // self::COLON_STATE_TEXT:
2212                                 switch( $c ) {
2213                                 case "<":
2214                                         // Could be either a <start> tag or an </end> tag
2215                                         $state = self::COLON_STATE_TAGSTART;
2216                                         break;
2217                                 case ":":
2218                                         if( $stack == 0 ) {
2219                                                 // We found it!
2220                                                 $before = substr( $str, 0, $i );
2221                                                 $after = substr( $str, $i + 1 );
2222                                                 wfProfileOut( $fname );
2223                                                 return $i;
2224                                         }
2225                                         // Embedded in a tag; don't break it.
2226                                         break;
2227                                 default:
2228                                         // Skip ahead looking for something interesting
2229                                         $colon = strpos( $str, ':', $i );
2230                                         if( $colon === false ) {
2231                                                 // Nothing else interesting
2232                                                 wfProfileOut( $fname );
2233                                                 return false;
2234                                         }
2235                                         $lt = strpos( $str, '<', $i );
2236                                         if( $stack === 0 ) {
2237                                                 if( $lt === false || $colon < $lt ) {
2238                                                         // We found it!
2239                                                         $before = substr( $str, 0, $colon );
2240                                                         $after = substr( $str, $colon + 1 );
2241                                                         wfProfileOut( $fname );
2242                                                         return $i;
2243                                                 }
2244                                         }
2245                                         if( $lt === false ) {
2246                                                 // Nothing else interesting to find; abort!
2247                                                 // We're nested, but there's no close tags left. Abort!
2248                                                 break 2;
2249                                         }
2250                                         // Skip ahead to next tag start
2251                                         $i = $lt;
2252                                         $state = self::COLON_STATE_TAGSTART;
2253                                 }
2254                                 break;
2255                         case 1: // self::COLON_STATE_TAG:
2256                                 // In a <tag>
2257                                 switch( $c ) {
2258                                 case ">":
2259                                         $stack++;
2260                                         $state = self::COLON_STATE_TEXT;
2261                                         break;
2262                                 case "/":
2263                                         // Slash may be followed by >?
2264                                         $state = self::COLON_STATE_TAGSLASH;
2265                                         break;
2266                                 default:
2267                                         // ignore
2268                                 }
2269                                 break;
2270                         case 2: // self::COLON_STATE_TAGSTART:
2271                                 switch( $c ) {
2272                                 case "/":
2273                                         $state = self::COLON_STATE_CLOSETAG;
2274                                         break;
2275                                 case "!":
2276                                         $state = self::COLON_STATE_COMMENT;
2277                                         break;
2278                                 case ">":
2279                                         // Illegal early close? This shouldn't happen D:
2280                                         $state = self::COLON_STATE_TEXT;
2281                                         break;
2282                                 default:
2283                                         $state = self::COLON_STATE_TAG;
2284                                 }
2285                                 break;
2286                         case 3: // self::COLON_STATE_CLOSETAG:
2287                                 // In a </tag>
2288                                 if( $c == ">" ) {
2289                                         $stack--;
2290                                         if( $stack < 0 ) {
2291                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2292                                                 wfProfileOut( $fname );
2293                                                 return false;
2294                                         }
2295                                         $state = self::COLON_STATE_TEXT;
2296                                 }
2297                                 break;
2298                         case self::COLON_STATE_TAGSLASH:
2299                                 if( $c == ">" ) {
2300                                         // Yes, a self-closed tag <blah/>
2301                                         $state = self::COLON_STATE_TEXT;
2302                                 } else {
2303                                         // Probably we're jumping the gun, and this is an attribute
2304                                         $state = self::COLON_STATE_TAG;
2305                                 }
2306                                 break;
2307                         case 5: // self::COLON_STATE_COMMENT:
2308                                 if( $c == "-" ) {
2309                                         $state = self::COLON_STATE_COMMENTDASH;
2310                                 }
2311                                 break;
2312                         case self::COLON_STATE_COMMENTDASH:
2313                                 if( $c == "-" ) {
2314                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2315                                 } else {
2316                                         $state = self::COLON_STATE_COMMENT;
2317                                 }
2318                                 break;
2319                         case self::COLON_STATE_COMMENTDASHDASH:
2320                                 if( $c == ">" ) {
2321                                         $state = self::COLON_STATE_TEXT;
2322                                 } else {
2323                                         $state = self::COLON_STATE_COMMENT;
2324                                 }
2325                                 break;
2326                         default:
2327                                 throw new MWException( "State machine error in $fname" );
2328                         }
2329                 }
2330                 if( $stack > 0 ) {
2331                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2332                         return false;
2333                 }
2334                 wfProfileOut( $fname );
2335                 return false;
2336         }
2337
2338         /**
2339          * Return value of a magic variable (like PAGENAME)
2340          *
2341          * @private
2342          */
2343         function getVariableValue( $index ) {
2344                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2345
2346                 /**
2347                  * Some of these require message or data lookups and can be
2348                  * expensive to check many times.
2349                  */
2350                 static $varCache = array();
2351                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2352                         if ( isset( $varCache[$index] ) ) {
2353                                 return $varCache[$index];
2354                         }
2355                 }
2356
2357                 $ts = time();
2358                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2359
2360                 # Use the time zone
2361                 global $wgLocaltimezone;
2362                 if ( isset( $wgLocaltimezone ) ) {
2363                         $oldtz = getenv( 'TZ' );
2364                         putenv( 'TZ='.$wgLocaltimezone );
2365                 }
2366
2367                 wfSuppressWarnings(); // E_STRICT system time bitching
2368                 $localTimestamp = date( 'YmdHis', $ts );
2369                 $localMonth = date( 'm', $ts );
2370                 $localMonthName = date( 'n', $ts );
2371                 $localDay = date( 'j', $ts );
2372                 $localDay2 = date( 'd', $ts );
2373                 $localDayOfWeek = date( 'w', $ts );
2374                 $localWeek = date( 'W', $ts );
2375                 $localYear = date( 'Y', $ts );
2376                 $localHour = date( 'H', $ts );
2377                 if ( isset( $wgLocaltimezone ) ) {
2378                         putenv( 'TZ='.$oldtz );
2379                 }
2380                 wfRestoreWarnings();
2381
2382                 switch ( $index ) {
2383                         case 'currentmonth':
2384                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2385                         case 'currentmonthname':
2386                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2387                         case 'currentmonthnamegen':
2388                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2389                         case 'currentmonthabbrev':
2390                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2391                         case 'currentday':
2392                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2393                         case 'currentday2':
2394                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2395                         case 'localmonth':
2396                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2397                         case 'localmonthname':
2398                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2399                         case 'localmonthnamegen':
2400                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2401                         case 'localmonthabbrev':
2402                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2403                         case 'localday':
2404                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2405                         case 'localday2':
2406                                 return $varCache[$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 $this->getRevisionTimestamp();
2479                         case 'namespace':
2480                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2481                         case 'namespacee':
2482                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2483                         case 'talkspace':
2484                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2485                         case 'talkspacee':
2486                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2487                         case 'subjectspace':
2488                                 return $this->mTitle->getSubjectNsText();
2489                         case 'subjectspacee':
2490                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2491                         case 'currentdayname':
2492                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2493                         case 'currentyear':
2494                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2495                         case 'currenttime':
2496                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2497                         case 'currenthour':
2498                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2499                         case 'currentweek':
2500                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2501                                 // int to remove the padding
2502                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2503                         case 'currentdow':
2504                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2505                         case 'localdayname':
2506                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2507                         case 'localyear':
2508                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2509                         case 'localtime':
2510                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2511                         case 'localhour':
2512                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2513                         case 'localweek':
2514                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2515                                 // int to remove the padding
2516                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2517                         case 'localdow':
2518                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2519                         case 'numberofarticles':
2520                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2521                         case 'numberoffiles':
2522                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2523                         case 'numberofusers':
2524                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2525                         case 'numberofpages':
2526                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2527                         case 'numberofadmins':
2528                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2529                         case 'numberofedits':
2530                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2531                         case 'currenttimestamp':
2532                                 return $varCache[$index] = wfTimestampNow();
2533                         case 'localtimestamp':
2534                                 return $varCache[$index] = $localTimestamp;
2535                         case 'currentversion':
2536                                 return $varCache[$index] = SpecialVersion::getVersion();
2537                         case 'sitename':
2538                                 return $wgSitename;
2539                         case 'server':
2540                                 return $wgServer;
2541                         case 'servername':
2542                                 return $wgServerName;
2543                         case 'scriptpath':
2544                                 return $wgScriptPath;
2545                         case 'directionmark':
2546                                 return $wgContLang->getDirMark();
2547                         case 'contentlanguage':
2548                                 global $wgContLanguageCode;
2549                                 return $wgContLanguageCode;
2550                         default:
2551                                 $ret = null;
2552                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2553                                         return $ret;
2554                                 else
2555                                         return null;
2556                 }
2557         }
2558
2559         /**
2560          * initialise the magic variables (like CURRENTMONTHNAME)
2561          *
2562          * @private
2563          */
2564         function initialiseVariables() {
2565                 $fname = 'Parser::initialiseVariables';
2566                 wfProfileIn( $fname );
2567                 $variableIDs = MagicWord::getVariableIDs();
2568
2569                 $this->mVariables = new MagicWordArray( $variableIDs );
2570                 wfProfileOut( $fname );
2571         }
2572
2573         /**
2574          * Preprocess some wikitext and return the document tree.
2575          * This is the ghost of replace_variables().
2576          *
2577          * @param string $text The text to parse
2578          * @param integer flags Bitwise combination of:
2579          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2580          *                                     included. Default is to assume a direct page view.
2581          *
2582          * The generated DOM tree must depend only on the input text, the flags, and $this->ot['msg'].
2583          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2584          *
2585          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2586          * change in the DOM tree for a given text, must be passed through the section identifier
2587          * in the section edit link and thus back to extractSections().
2588          *
2589          * The output of this function is currently only cached in process memory, but a persistent
2590          * cache may be implemented at a later date which takes further advantage of these strict
2591          * dependency requirements.
2592          *
2593          * @private
2594          */
2595         function preprocessToDom ( $text, $flags = 0 ) {
2596                 wfProfileIn( __METHOD__ );
2597                 wfProfileIn( __METHOD__.'-makexml' );
2598
2599                 static $msgRules, $normalRules, $inclusionSupertags, $nonInclusionSupertags;
2600                 if ( !$msgRules ) {
2601                         $msgRules = array(
2602                                 '{' => array(
2603                                         'end' => '}',
2604                                         'names' => array(
2605                                                 2 => 'template',
2606                                         ),
2607                                         'min' => 2,
2608                                         'max' => 2,
2609                                 ),
2610                                 '[' => array(
2611                                         'end' => ']',
2612                                         'names' => array( 2 => null ),
2613                                         'min' => 2,
2614                                         'max' => 2,
2615                                 )
2616                         );
2617                         $normalRules = array(
2618                                 '{' => array(
2619                                         'end' => '}',
2620                                         'names' => array(
2621                                                 2 => 'template',
2622                                                 3 => 'tplarg',
2623                                         ),
2624                                         'min' => 2,
2625                                         'max' => 3,
2626                                 ),
2627                                 '[' => array(
2628                                         'end' => ']',
2629                                         'names' => array( 2 => null ),
2630                                         'min' => 2,
2631                                         'max' => 2,
2632                                 )
2633                         );
2634                 }
2635                 if ( $this->ot['msg'] ) {
2636                         $rules = $msgRules;
2637                 } else {
2638                         $rules = $normalRules;
2639                 }
2640                 $forInclusion = $flags & self::PTD_FOR_INCLUSION;
2641
2642                 $xmlishElements = $this->getStripList();
2643                 $enableOnlyinclude = false;
2644                 if ( $forInclusion ) {
2645                         $ignoredTags = array( 'includeonly', '/includeonly' );
2646                         $ignoredElements = array( 'noinclude' );
2647                         $xmlishElements[] = 'noinclude';
2648                         if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
2649                                 $enableOnlyinclude = true;
2650                         }
2651                 } else {
2652                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
2653                         $ignoredElements = array( 'includeonly' );
2654                         $xmlishElements[] = 'includeonly';
2655                 }
2656                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
2657
2658                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2659                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
2660
2661                 $stack = array();      # Stack of unclosed parentheses
2662                 $stackIndex = -1;      # Stack read pointer
2663
2664                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2665                 $revText = strrev( $text ); // For fast reverse searches
2666
2667                 $i = 0;                     # Input pointer, starts out pointing to a pseudo-newline before the start
2668                 $topAccum = '<root>';      # Top level text accumulator
2669                 $accum =& $topAccum;            # Current text accumulator
2670                 $findEquals = false;            # True to find equals signs in arguments
2671                 $findHeading = false;           # True to look at LF characters for possible headings
2672                 $findPipe = false;              # True to take notice of pipe characters
2673                 $headingIndex = 1;
2674                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2675                 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
2676                 $fakeLineStart = true;     # Do a line-start run without outputting an LF character
2677
2678                 while ( true ) {
2679                         if ( $findOnlyinclude ) {
2680                                 // Ignore all input up to the next <onlyinclude>
2681                                 $startPos = strpos( $text, '<onlyinclude>', $i );
2682                                 if ( $startPos === false ) {
2683                                         // Ignored section runs to the end
2684                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
2685                                         break;
2686                                 }
2687                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
2688                                 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
2689                                 $i = $tagEndPos;
2690                                 $findOnlyinclude = false;
2691                         }
2692
2693                         if ( $fakeLineStart ) {
2694                                 $found = 'line-start';
2695                                 $curChar = '';
2696                         } else {
2697                                 # Find next opening brace, closing brace or pipe
2698                                 $search = $searchBase;
2699                                 if ( $stackIndex == -1 ) {
2700                                         $currentClosing = '';
2701                                         // Look for headings only at the top stack level
2702                                         // Among other things, this resolves the ambiguity between =
2703                                         // for headings and = for template arguments
2704                                         $search .= "\n";
2705                                 } else {
2706                                         $currentClosing = $stack[$stackIndex]['close'];
2707                                         $search .= $currentClosing;
2708                                 }
2709                                 if ( $findPipe ) {
2710                                         $search .= '|';
2711                                 }
2712                                 if ( $findEquals ) {
2713                                         $search .= '=';
2714                                 }
2715                                 $rule = null;
2716                                 # Output literal section, advance input counter
2717                                 $literalLength = strcspn( $text, $search, $i );
2718                                 if ( $literalLength > 0 ) {
2719                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2720                                         $i += $literalLength;
2721                                 }
2722                                 if ( $i >= strlen( $text ) ) {
2723                                         if ( $currentClosing == "\n" ) {
2724                                                 // Do a past-the-end run to finish off the heading
2725                                                 $curChar = '';
2726                                                 $found = 'line-end';
2727                                         } else {
2728                                                 # All done
2729                                                 break;
2730                                         }
2731                                 } else {
2732                                         $curChar = $text[$i];
2733                                         if ( $curChar == '|' ) {
2734                                                 $found = 'pipe';
2735                                         } elseif ( $curChar == '=' ) {
2736                                                 $found = 'equals';
2737                                         } elseif ( $curChar == '<' ) {
2738                                                 $found = 'angle';
2739                                         } elseif ( $curChar == "\n" ) {
2740                                                 if ( $stackIndex == -1 ) {
2741                                                         $found = 'line-start';
2742                                                 } else {
2743                                                         $found = 'line-end';
2744                                                 }
2745                                         } elseif ( $curChar == $currentClosing ) {
2746                                                 $found = 'close';
2747                                         } elseif ( isset( $rules[$curChar] ) ) {
2748                                                 $found = 'open';
2749                                                 $rule = $rules[$curChar];
2750                                         } else {
2751                                                 # Some versions of PHP have a strcspn which stops on null characters
2752                                                 # Ignore and continue
2753                                                 ++$i;
2754                                                 continue;
2755                                         }
2756                                 }
2757                         }
2758
2759                         if ( $found == 'angle' ) {
2760                                 $matches = false;
2761                                 // Handle </onlyinclude>
2762                                 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
2763                                         $findOnlyinclude = true;
2764                                         continue;
2765                                 }
2766
2767                                 // Determine element name
2768                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
2769                                         // Element name missing or not listed
2770                                         $accum .= '&lt;';
2771                                         ++$i;
2772                                         continue;
2773                                 }
2774                                 // Handle comments
2775                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2776                                         // To avoid leaving blank lines, when a comment is both preceded
2777                                         // and followed by a newline (ignoring spaces), trim leading and
2778                                         // trailing spaces and one of the newlines.
2779
2780                                         // Find the end
2781                                         $endPos = strpos( $text, '-->', $i + 4 );
2782                                         if ( $endPos === false ) {
2783                                                 // Unclosed comment in input, runs to end
2784                                                 $inner = substr( $text, $i );
2785                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2786                                                 $i = strlen( $text );
2787                                         } else {
2788                                                 // Search backwards for leading whitespace
2789                                                 $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
2790                                                 // Search forwards for trailing whitespace
2791                                                 // $wsEnd will be the position of the last space
2792                                                 $wsEnd = $endPos + 2 + strspn( $text, ' ', $endPos + 3 );
2793                                                 // Eat the line if possible
2794                                                 // TODO: This could theoretically be done if $wsStart == 0, i.e. for comments at
2795                                                 // the overall start. That's not how Sanitizer::removeHTMLcomments() does it, but
2796                                                 // it's a possible beneficial b/c break.
2797                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
2798                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n" )
2799                                                 {
2800                                                         $startPos = $wsStart;
2801                                                         $endPos = $wsEnd + 1;
2802                                                         // Remove leading whitespace from the end of the accumulator
2803                                                         // Sanity check first though
2804                                                         $wsLength = $i - $wsStart;
2805                                                         if ( $wsLength > 0 && substr( $accum, -$wsLength ) === str_repeat( ' ', $wsLength ) ) {
2806                                                                 $accum = substr( $accum, 0, -$wsLength );
2807                                                         }
2808                                                         // Do a line-start run next time to look for headings after the comment,
2809                                                         // but only if stackIndex=-1, because headings don't exist at deeper levels.
2810                                                         if ( $stackIndex == -1 ) {
2811                                                                 $fakeLineStart = true;
2812                                                         }
2813                                                 } else {
2814                                                         // No line to eat, just take the comment itself
2815                                                         $startPos = $i;
2816                                                         $endPos += 2;
2817                                                 }
2818
2819                                                 $i = $endPos + 1;
2820                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
2821                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2822                                         }
2823                                         continue;
2824                                 }
2825                                 $name = $matches[1];
2826                                 $attrStart = $i + strlen( $name ) + 1;
2827
2828                                 // Find end of tag
2829                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2830                                 if ( $tagEndPos === false ) {
2831                                         // Infinite backtrack
2832                                         // Disable tag search to prevent worst-case O(N^2) performance
2833                                         $noMoreGT = true;
2834                                         $accum .= '&lt;';
2835                                         ++$i;
2836                                         continue;
2837                                 }
2838
2839                                 // Handle ignored tags
2840                                 if ( in_array( $name, $ignoredTags ) ) {
2841                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '</ignore>';
2842                                         $i = $tagEndPos + 1;
2843                                         continue;
2844                                 }
2845
2846                                 $tagStartPos = $i;
2847                                 if ( $text[$tagEndPos-1] == '/' ) {
2848                                         $attrEnd = $tagEndPos - 1;
2849                                         $inner = null;
2850                                         $i = $tagEndPos + 1;
2851                                         $close = '';
2852                                 } else {
2853                                         $attrEnd = $tagEndPos;
2854                                         // Find closing tag
2855                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2856                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2857                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2858                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2859                                         } else {
2860                                                 // No end tag -- let it run out to the end of the text.
2861                                                 $inner = substr( $text, $tagEndPos + 1 );
2862                                                 $i = strlen( $text );
2863                                                 $close = '';
2864                                         }
2865                                 }
2866                                 // <includeonly> and <noinclude> just become <ignore> tags
2867                                 if ( in_array( $name, $ignoredElements ) ) {
2868                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
2869                                                 . '</ignore>';
2870                                         continue;
2871                                 }
2872
2873                                 $accum .= '<ext>';
2874                                 if ( $attrEnd <= $attrStart ) {
2875                                         $attr = '';
2876                                 } else {
2877                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2878                                 }
2879                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2880                                         // Note that the attr element contains the whitespace between name and attribute,
2881                                         // this is necessary for precise reconstruction during pre-save transform.
2882                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2883                                 if ( $inner !== null ) {
2884                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2885                                 }
2886                                 $accum .= $close . '</ext>';
2887                         }
2888
2889                         elseif ( $found == 'line-start' ) {
2890                                 // Is this the start of a heading?
2891                                 // Line break belongs before the heading element in any case
2892                                 if ( $fakeLineStart ) {
2893                                         $fakeLineStart = false;
2894                                 } else {
2895                                         $accum .= $curChar;
2896                                         $i++;
2897                                 }
2898
2899                                 $count = strspn( $text, '=', $i, 6 );
2900                                 if ( $count > 0 ) {
2901                                         $piece = array(
2902                                                 'open' => "\n",
2903                                                 'close' => "\n",
2904                                                 'parts' => array( str_repeat( '=', $count ) ),
2905                                                 'startPos' => $i,
2906                                                 'count' => $count );
2907                                         $stack[++$stackIndex] = $piece;
2908                                         $i += $count;
2909                                         $accum =& $stack[$stackIndex]['parts'][0];
2910                                         $findPipe = false;
2911                                 }
2912                         }
2913
2914                         elseif ( $found == 'line-end' ) {
2915                                 $piece = $stack[$stackIndex];
2916                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2917                                 assert( $piece['open'] == "\n" );
2918                                 assert( $stackIndex == 0 );
2919                                 // Search back through the input to see if it has a proper close
2920                                 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
2921                                 $m = false;
2922                                 $count = $piece['count'];
2923                                 if ( preg_match( "/\s*(={{$count}})/A", $revText, $m, 0, strlen( $text ) - $i ) ) {
2924                                         if ( $i - strlen( $m[0] ) == $piece['startPos'] ) {
2925                                                 // This is just a single string of equals signs on its own line
2926                                                 // Divide by two and round down to create start and end delimiters
2927                                                 $count = intval( $count / 2 );
2928                                         } else {
2929                                                 $count = min( strlen( $m[1] ), $count );
2930                                         }
2931                                         if ( $count > 0 ) {
2932                                                 // Normal match, output <h>
2933                                                 $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2934                                                 $headingIndex++;
2935                                         } else {
2936                                                 // Single equals sign on its own line, count=0
2937                                                 $element = $accum;
2938                                         }
2939                                 } else {
2940                                         // No match, no <h>, just pass down the inner text
2941                                         $element = $accum;
2942                                 }
2943                                 // Unwind the stack
2944                                 // Headings can only occur on the top level, so this is a bit simpler than the
2945                                 // generic stack unwind operation in the close case
2946                                 unset( $stack[$stackIndex--] );
2947                                 $accum =& $topAccum;
2948                                 $findEquals = false;
2949                                 $findPipe = false;
2950
2951                                 // Append the result to the enclosing accumulator
2952                                 $accum .= $element;
2953                                 // Note that we do NOT increment the input pointer.
2954                                 // This is because the closing linebreak could be the opening linebreak of
2955                                 // another heading. Infinite loops are avoided because the next iteration MUST
2956                                 // hit the heading open case above, which unconditionally increments the
2957                                 // input pointer.
2958                         }
2959
2960                         elseif ( $found == 'open' ) {
2961                                 # count opening brace characters
2962                                 $count = strspn( $text, $curChar, $i );
2963
2964                                 # we need to add to stack only if opening brace count is enough for one of the rules
2965                                 if ( $count >= $rule['min'] ) {
2966                                         # Add it to the stack
2967                                         $piece = array(
2968                                                 'open' => $curChar,
2969                                                 'close' => $rule['end'],
2970                                                 'count' => $count,
2971                                                 'parts' => array( '' ),
2972                                                 'eqpos' => array(),
2973                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2974                                         );
2975
2976                                         $stackIndex ++;
2977                                         $stack[$stackIndex] = $piece;
2978                                         $accum =& $stack[$stackIndex]['parts'][0];
2979                                         $findEquals = false;
2980                                         $findPipe = true;
2981                                 } else {
2982                                         # Add literal brace(s)
2983                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2984                                 }
2985                                 $i += $count;
2986                         }
2987
2988                         elseif ( $found == 'close' ) {
2989                                 $piece = $stack[$stackIndex];
2990                                 # lets check if there are enough characters for closing brace
2991                                 $maxCount = $piece['count'];
2992                                 $count = strspn( $text, $curChar, $i, $maxCount );
2993
2994                                 # check for maximum matching characters (if there are 5 closing
2995                                 # characters, we will probably need only 3 - depending on the rules)
2996                                 $matchingCount = 0;
2997                                 $rule = $rules[$piece['open']];
2998                                 if ( $count > $rule['max'] ) {
2999                                         # The specified maximum exists in the callback array, unless the caller
3000                                         # has made an error
3001                                         $matchingCount = $rule['max'];
3002                                 } else {
3003                                         # Count is less than the maximum
3004                                         # Skip any gaps in the callback array to find the true largest match
3005                                         # Need to use array_key_exists not isset because the callback can be null
3006                                         $matchingCount = $count;
3007                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
3008                                                 --$matchingCount;
3009                                         }
3010                                 }
3011
3012                                 if ($matchingCount <= 0) {
3013                                         # No matching element found in callback array
3014                                         # Output a literal closing brace and continue
3015                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
3016                                         $i += $count;
3017                                         continue;
3018                                 }
3019                                 $name = $rule['names'][$matchingCount];
3020                                 if ( $name === null ) {
3021                                         // No element, just literal text
3022                                         $element = str_repeat( $piece['open'], $matchingCount ) .
3023                                                 implode( '|', $piece['parts'] ) .
3024                                                 str_repeat( $rule['end'], $matchingCount );
3025                                 } else {
3026                                         # Create XML element
3027                                         # Note: $parts is already XML, does not need to be encoded further
3028                                         $parts = $piece['parts'];
3029                                         $title = $parts[0];
3030                                         unset( $parts[0] );
3031
3032                                         # The invocation is at the start of the line if lineStart is set in
3033                                         # the stack, and all opening brackets are used up.
3034                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
3035                                                 $attr = ' lineStart="1"';
3036                                         } else {
3037                                                 $attr = '';
3038                                         }
3039
3040                                         $element = "<$name$attr>";
3041                                         $element .= "<title>$title</title>";
3042                                         $argIndex = 1;
3043                                         foreach ( $parts as $partIndex => $part ) {
3044                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
3045                                                         $eqpos = $piece['eqpos'][$partIndex];
3046                                                         $argName = substr( $part, 0, $eqpos );
3047                                                         $argValue = substr( $part, $eqpos + 1 );
3048                                                         $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
3049                                                 } else {
3050                                                         $element .= "<part><name index=\"$argIndex\" /><value>$part</value></part>";
3051                                                         $argIndex++;
3052                                                 }
3053                                         }
3054                                         $element .= "</$name>";
3055                                 }
3056
3057                                 # Advance input pointer
3058                                 $i += $matchingCount;
3059
3060                                 # Unwind the stack
3061                                 unset( $stack[$stackIndex--] );
3062                                 if ( $stackIndex == -1 ) {
3063                                         $accum =& $topAccum;
3064                                         $findEquals = false;
3065                                         $findPipe = false;
3066                                 } else {
3067                                         $partCount = count( $stack[$stackIndex]['parts'] );
3068                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
3069                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
3070                                         $findEquals = $findPipe && $partCount > 1
3071                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
3072                                 }
3073
3074                                 # Re-add the old stack element if it still has unmatched opening characters remaining
3075                                 if ($matchingCount < $piece['count']) {
3076                                         $piece['parts'] = array( '' );
3077                                         $piece['count'] -= $matchingCount;
3078                                         $piece['eqpos'] = array();
3079                                         # do we still qualify for any callback with remaining count?
3080                                         $names = $rules[$piece['open']]['names'];
3081                                         $skippedBraces = 0;
3082                                         $enclosingAccum =& $accum;
3083                                         while ( $piece['count'] ) {
3084                                                 if ( array_key_exists( $piece['count'], $names ) ) {
3085                                                         $stackIndex++;
3086                                                         $stack[$stackIndex] = $piece;
3087                                                         $accum =& $stack[$stackIndex]['parts'][0];
3088                                                         $findEquals = true;
3089                                                         $findPipe = true;
3090                                                         break;
3091                                                 }
3092                                                 --$piece['count'];
3093                                                 $skippedBraces ++;
3094                                         }
3095                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
3096                                 }
3097
3098                                 # Add XML element to the enclosing accumulator
3099                                 $accum .= $element;
3100                         }
3101
3102                         elseif ( $found == 'pipe' ) {
3103                                 $stack[$stackIndex]['parts'][] = '';
3104                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3105                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
3106                                 $findEquals = true;
3107                                 ++$i;
3108                         }
3109
3110                         elseif ( $found == 'equals' ) {
3111                                 $findEquals = false;
3112                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3113                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
3114                                 $accum .= '=';
3115                                 ++$i;
3116                         }
3117                 }
3118
3119                 # Output any remaining unclosed brackets
3120                 foreach ( $stack as $piece ) {
3121                         if ( $piece['open'] == "\n" ) {
3122                                 $topAccum .= $piece['parts'][0];
3123                         } else {
3124                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
3125                         }
3126                 }
3127                 $topAccum .= '</root>';
3128
3129                 wfProfileOut( __METHOD__.'-makexml' );
3130                 wfProfileIn( __METHOD__.'-loadXML' );
3131                 $dom = new DOMDocument;
3132                 wfSuppressWarnings();
3133                 $result = $dom->loadXML( $topAccum );
3134                 wfRestoreWarnings();
3135                 if ( !$result ) {
3136                         // Try running the XML through UtfNormal to get rid of invalid characters
3137                         $topAccum = UtfNormal::cleanUp( $topAccum );
3138                         $result = $dom->loadXML( $topAccum );
3139                         if ( !$result ) {
3140                                 throw new MWException( __METHOD__.' generated invalid XML' );
3141                         }
3142                 }
3143                 wfProfileOut( __METHOD__.'-loadXML' );
3144                 wfProfileOut( __METHOD__ );
3145                 return $dom;
3146         }
3147
3148         /*
3149          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3150          */
3151         public static function splitWhitespace( $s ) {
3152                 $ltrimmed = ltrim( $s );
3153                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3154                 $trimmed = rtrim( $ltrimmed );
3155                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3156                 if ( $diff > 0 ) {
3157                         $w2 = substr( $ltrimmed, -$diff );
3158                 } else {
3159                         $w2 = '';
3160                 }
3161                 return array( $w1, $trimmed, $w2 );
3162         }
3163
3164         /**
3165          * Replace magic variables, templates, and template arguments
3166          * with the appropriate text. Templates are substituted recursively,
3167          * taking care to avoid infinite loops.
3168          *
3169          * Note that the substitution depends on value of $mOutputType:
3170          *  OT_WIKI: only {{subst:}} templates
3171          *  OT_MSG: only magic variables
3172          *  OT_HTML: all templates and magic variables
3173          *
3174          * @param string $tex The text to transform
3175          * @param PPFrame $frame Object describing the arguments passed to the template
3176          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3177          * @private
3178          */
3179         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3180                 # Prevent too big inclusions
3181                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3182                         return $text;
3183                 }
3184
3185                 $fname = __METHOD__;
3186                 wfProfileIn( $fname );
3187
3188                 if ( $frame === false ) {
3189                         $frame = new PPFrame( $this );
3190                 } elseif ( !( $frame instanceof PPFrame ) ) {
3191                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3192                 }
3193
3194                 $dom = $this->preprocessToDom( $text );
3195                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3196                 $text = $frame->expand( $dom, $flags );
3197
3198                 wfProfileOut( $fname );
3199                 return $text;
3200         }
3201
3202         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3203         static function createAssocArgs( $args ) {
3204                 $assocArgs = array();
3205                 $index = 1;
3206                 foreach( $args as $arg ) {
3207                         $eqpos = strpos( $arg, '=' );
3208                         if ( $eqpos === false ) {
3209                                 $assocArgs[$index++] = $arg;
3210                         } else {
3211                                 $name = trim( substr( $arg, 0, $eqpos ) );
3212                                 $value = trim( substr( $arg, $eqpos+1 ) );
3213                                 if ( $value === false ) {
3214                                         $value = '';
3215                                 }
3216                                 if ( $name !== false ) {
3217                                         $assocArgs[$name] = $value;
3218                                 }
3219                         }
3220                 }
3221
3222                 return $assocArgs;
3223         }
3224
3225         /**
3226          * Return the text of a template, after recursively
3227          * replacing any variables or templates within the template.
3228          *
3229          * @param array $piece The parts of the template
3230          *  $piece['text']: matched text
3231          *  $piece['title']: the title, i.e. the part before the |
3232          *  $piece['parts']: the parameter array
3233          * @param PPFrame The current frame, contains template arguments
3234          * @return string the text of the template
3235          * @private
3236          */
3237         function braceSubstitution( $piece, $frame ) {
3238                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3239                 $fname = __METHOD__;
3240                 wfProfileIn( $fname );
3241                 wfProfileIn( __METHOD__.'-setup' );
3242
3243                 # Flags
3244                 $found = false;             # $text has been filled
3245                 $nowiki = false;            # wiki markup in $text should be escaped
3246                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3247                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3248                 $isDOM = false;             # $text is a DOM node needing expansion
3249
3250                 # Title object, where $text came from
3251                 $title = NULL;
3252
3253                 # $part1 is the bit before the first |, and must contain only title characters.
3254                 # Various prefixes will be stripped from it later.
3255                 $titleWithSpaces = $frame->expand( $piece['title'] );
3256                 $part1 = trim( $titleWithSpaces );
3257                 $titleText = false;
3258
3259                 # Original title text preserved for various purposes
3260                 $originalTitle = $part1;
3261
3262                 # $args is a list of argument nodes, starting from index 0, not including $part1
3263                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3264                 wfProfileOut( __METHOD__.'-setup' );
3265
3266                 # SUBST
3267                 wfProfileIn( __METHOD__.'-modifiers' );
3268                 if ( !$found ) {
3269                         $mwSubst =& MagicWord::get( 'subst' );
3270                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3271                                 # One of two possibilities is true:
3272                                 # 1) Found SUBST but not in the PST phase
3273                                 # 2) Didn't find SUBST and in the PST phase
3274                                 # In either case, return without further processing
3275                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3276                                 $found = true;
3277                         }
3278                 }
3279
3280                 # Variables
3281                 if ( !$found && $args->length == 0 ) {
3282                         $id = $this->mVariables->matchStartToEnd( $part1 );
3283                         if ( $id !== false ) {
3284                                 $text = $this->getVariableValue( $id );
3285                                 if (MagicWord::getCacheTTL($id)>-1)
3286                                         $this->mOutput->mContainsOldMagic = true;
3287                                 $found = true;
3288                         }
3289                 }
3290
3291                 # MSG, MSGNW and RAW
3292                 if ( !$found ) {
3293                         # Check for MSGNW:
3294                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3295                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3296                                 $nowiki = true;
3297                         } else {
3298                                 # Remove obsolete MSG:
3299                                 $mwMsg =& MagicWord::get( 'msg' );
3300                                 $mwMsg->matchStartAndRemove( $part1 );
3301                         }
3302
3303                         # Check for RAW:
3304                         $mwRaw =& MagicWord::get( 'raw' );
3305                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3306                                 $forceRawInterwiki = true;
3307                         }
3308                 }
3309                 wfProfileOut( __METHOD__.'-modifiers' );
3310
3311                 # Parser functions
3312                 if ( !$found ) {
3313                         wfProfileIn( __METHOD__ . '-pfunc' );
3314
3315                         $colonPos = strpos( $part1, ':' );
3316                         if ( $colonPos !== false ) {
3317                                 # Case sensitive functions
3318                                 $function = substr( $part1, 0, $colonPos );
3319                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3320                                         $function = $this->mFunctionSynonyms[1][$function];
3321                                 } else {
3322                                         # Case insensitive functions
3323                                         $function = strtolower( $function );
3324                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3325                                                 $function = $this->mFunctionSynonyms[0][$function];
3326                                         } else {
3327                                                 $function = false;
3328                                         }
3329                                 }
3330                                 if ( $function ) {
3331                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3332                                         $initialArgs = array( &$this );
3333                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3334                                         if ( $flags & SFH_OBJECT_ARGS ) {
3335                                                 # Add a frame parameter, and pass the arguments as an array
3336                                                 $allArgs = $initialArgs;
3337                                                 $allArgs[] = $frame;
3338                                                 foreach ( $args as $arg ) {
3339                                                         $funcArgs[] = $arg;
3340                                                 }
3341                                                 $allArgs[] = $funcArgs;
3342                                         } else {
3343                                                 # Convert arguments to plain text
3344                                                 foreach ( $args as $arg ) {
3345                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3346                                                 }
3347                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3348                                         }
3349
3350                                         $result = call_user_func_array( $callback, $allArgs );
3351                                         $found = true;
3352
3353                                         if ( is_array( $result ) ) {
3354                                                 if ( isset( $result[0] ) ) {
3355                                                         $text = $result[0];
3356                                                         unset( $result[0] );
3357                                                 }
3358
3359                                                 // Extract flags into the local scope
3360                                                 // This allows callers to set flags such as nowiki, found, etc.
3361                                                 extract( $result );
3362                                         } else {
3363                                                 $text = $result;
3364                                         }
3365                                 }
3366                         }
3367                         wfProfileOut( __METHOD__ . '-pfunc' );
3368                 }
3369
3370                 # Finish mangling title and then check for loops.
3371                 # Set $title to a Title object and $titleText to the PDBK
3372                 if ( !$found ) {
3373                         $ns = NS_TEMPLATE;
3374                         # Split the title into page and subpage
3375                         $subpage = '';
3376                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3377                         if ($subpage !== '') {
3378                                 $ns = $this->mTitle->getNamespace();
3379                         }
3380                         $title = Title::newFromText( $part1, $ns );
3381                         if ( $title ) {
3382                                 $titleText = $title->getPrefixedText();
3383                                 # Check for language variants if the template is not found
3384                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3385                                         $wgContLang->findVariantLink($part1, $title);
3386                                 }
3387                                 # Do infinite loop check
3388                                 if ( !$frame->loopCheck( $title ) ) {
3389                                         $found = true;
3390                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
3391                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3392                                 }
3393                                 # Do recursion depth check
3394                                 $limit = $this->mOptions->getMaxTemplateDepth();
3395                                 if ( $frame->depth >= $limit ) {
3396                                         $found = true;
3397                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
3398                                 }
3399                         }
3400                 }
3401
3402                 # Load from database
3403                 if ( !$found && $title ) {
3404                         wfProfileIn( __METHOD__ . '-loadtpl' );
3405                         if ( !$title->isExternal() ) {
3406                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3407                                         $text = SpecialPage::capturePath( $title );
3408                                         if ( is_string( $text ) ) {
3409                                                 $found = true;
3410                                                 $isHTML = true;
3411                                                 $this->disableCache();
3412                                         }
3413                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3414                                         $found = false; //access denied
3415                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3416                                 } else {
3417                                         list( $text, $title ) = $this->getTemplateDom( $title );
3418                                         if ( $text !== false ) {
3419                                                 $found = true;
3420                                                 $isDOM = true;
3421                                         }
3422                                 }
3423
3424                                 # If the title is valid but undisplayable, make a link to it
3425                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3426                                         $text = "[[:$titleText]]";
3427                                         $found = true;
3428                                 }
3429                         } elseif ( $title->isTrans() ) {
3430                                 // Interwiki transclusion
3431                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3432                                         $text = $this->interwikiTransclude( $title, 'render' );
3433                                         $isHTML = true;
3434                                 } else {
3435                                         $text = $this->interwikiTransclude( $title, 'raw' );
3436                                         // Preprocess it like a template
3437                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3438                                         $isDOM = true;
3439                                 }
3440                                 $found = true;
3441                         }
3442                         wfProfileOut( __METHOD__ . '-loadtpl' );
3443                 }
3444
3445                 # If we haven't found text to substitute by now, we're done
3446                 # Recover the source wikitext and return it
3447                 if ( !$found ) {
3448                         $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3449                         wfProfileOut( $fname );
3450                         return $text;
3451                 }
3452
3453                 # Expand DOM-style return values in a child frame
3454                 if ( $isDOM ) {
3455                         # Clean up argument array
3456                         $newFrame = $frame->newChild( $args, $title );
3457
3458                         if ( $titleText !== false && $newFrame->isEmpty() ) {
3459                                 # Expansion is eligible for the empty-frame cache
3460                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3461                                         $text = $this->mTplExpandCache[$titleText];
3462                                 } else {
3463                                         $text = $newFrame->expand( $text );
3464                                         $this->mTplExpandCache[$titleText] = $text;
3465                                 }
3466                         } else {
3467                                 # Uncached expansion
3468                                 $text = $newFrame->expand( $text );
3469                         }
3470                 }
3471
3472                 # Replace raw HTML by a placeholder
3473                 # Add a blank line preceding, to prevent it from mucking up
3474                 # immediately preceding headings
3475                 if ( $isHTML ) {
3476                         $text = "\n\n" . $this->insertStripItem( $text );
3477                 }
3478                 # Escape nowiki-style return values
3479                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3480                         $text = wfEscapeWikiText( $text );
3481                 }
3482                 # Bug 529: if the template begins with a table or block-level
3483                 # element, it should be treated as beginning a new line.
3484                 # This behaviour is somewhat controversial.
3485                 elseif ( !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3486                         $text = "\n" . $text;
3487                 }
3488
3489                 if ( !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3490                         # Error, oversize inclusion
3491                         $text = "[[$originalTitle]]" .
3492                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3493                 }
3494
3495                 wfProfileOut( $fname );
3496                 return $text;
3497         }
3498
3499         /**
3500          * Get the semi-parsed DOM representation of a template with a given title,
3501          * and its redirect destination title. Cached.
3502          */
3503         function getTemplateDom( $title ) {
3504                 $cacheTitle = $title;
3505                 $titleText = $title->getPrefixedDBkey();
3506
3507                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3508                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3509                         $title = Title::makeTitle( $ns, $dbk );
3510                         $titleText = $title->getPrefixedDBkey();
3511                 }
3512                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3513                         return array( $this->mTplDomCache[$titleText], $title );
3514                 }
3515
3516                 // Cache miss, go to the database
3517                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3518
3519                 if ( $text === false ) {
3520                         $this->mTplDomCache[$titleText] = false;
3521                         return array( false, $title );
3522                 }
3523
3524                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3525                 $this->mTplDomCache[ $titleText ] = $dom;
3526
3527                 if (! $title->equals($cacheTitle)) {
3528                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3529                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3530                 }
3531
3532                 return array( $dom, $title );
3533         }
3534
3535         /**
3536          * Fetch the unparsed text of a template and register a reference to it.
3537          */
3538         function fetchTemplateAndTitle( $title ) {
3539                 $templateCb = $this->mOptions->getTemplateCallback();
3540                 $stuff = call_user_func( $templateCb, $title );
3541                 $text = $stuff['text'];
3542                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3543                 if ( isset( $stuff['deps'] ) ) {
3544                         foreach ( $stuff['deps'] as $dep ) {
3545                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3546                         }
3547                 }
3548                 return array($text,$finalTitle);
3549         }
3550
3551         function fetchTemplate( $title ) {
3552                 $rv = $this->fetchTemplateAndTitle($title);
3553                 return $rv[0];
3554         }
3555
3556         /**
3557          * Static function to get a template
3558          * Can be overridden via ParserOptions::setTemplateCallback().
3559          */
3560         static function statelessFetchTemplate( $title ) {
3561                 $text = $skip = false;
3562                 $finalTitle = $title;
3563                 $deps = array();
3564
3565                 // Loop to fetch the article, with up to 1 redirect
3566                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3567                         # Give extensions a chance to select the revision instead
3568                         $id = false; // Assume current
3569                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3570
3571                         if( $skip ) {
3572                                 $text = false;
3573                                 $deps[] = array(
3574                                         'title' => $title,
3575                                         'page_id' => $title->getArticleID(),
3576                                         'rev_id' => null );
3577                                 break;
3578                         }
3579                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3580                         $rev_id = $rev ? $rev->getId() : 0;
3581
3582                         $deps[] = array(
3583                                 'title' => $title,
3584                                 'page_id' => $title->getArticleID(),
3585                                 'rev_id' => $rev_id );
3586
3587                         if( $rev ) {
3588                                 $text = $rev->getText();
3589                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3590                                 global $wgLang;
3591                                 $message = $wgLang->lcfirst( $title->getText() );
3592                                 $text = wfMsgForContentNoTrans( $message );
3593                                 if( wfEmptyMsg( $message, $text ) ) {
3594                                         $text = false;
3595                                         break;
3596                                 }
3597                         } else {
3598                                 break;
3599                         }
3600                         if ( $text === false ) {
3601                                 break;
3602                         }
3603                         // Redirect?
3604                         $finalTitle = $title;
3605                         $title = Title::newFromRedirect( $text );
3606                 }
3607                 return array(
3608                         'text' => $text,
3609                         'finalTitle' => $finalTitle,
3610                         'deps' => $deps );
3611         }
3612
3613         /**
3614          * Transclude an interwiki link.
3615          */
3616         function interwikiTransclude( $title, $action ) {
3617                 global $wgEnableScaryTranscluding;
3618
3619                 if (!$wgEnableScaryTranscluding)
3620                         return wfMsg('scarytranscludedisabled');
3621
3622                 $url = $title->getFullUrl( "action=$action" );
3623
3624                 if (strlen($url) > 255)
3625                         return wfMsg('scarytranscludetoolong');
3626                 return $this->fetchScaryTemplateMaybeFromCache($url);
3627         }
3628
3629         function fetchScaryTemplateMaybeFromCache($url) {
3630                 global $wgTranscludeCacheExpiry;
3631                 $dbr = wfGetDB(DB_SLAVE);
3632                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3633                                 array('tc_url' => $url));
3634                 if ($obj) {
3635                         $time = $obj->tc_time;
3636                         $text = $obj->tc_contents;
3637                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3638                                 return $text;
3639                         }
3640                 }
3641
3642                 $text = Http::get($url);
3643                 if (!$text)
3644                         return wfMsg('scarytranscludefailed', $url);
3645
3646                 $dbw = wfGetDB(DB_MASTER);
3647                 $dbw->replace('transcache', array('tc_url'), array(
3648                         'tc_url' => $url,
3649                         'tc_time' => time(),
3650                         'tc_contents' => $text));
3651                 return $text;
3652         }
3653
3654
3655         /**
3656          * Triple brace replacement -- used for template arguments
3657          * @private
3658          */
3659         function argSubstitution( $piece, $frame ) {
3660                 wfProfileIn( __METHOD__ );
3661
3662                 $error = false;
3663                 $parts = $piece['parts'];
3664                 $nameWithSpaces = $frame->expand( $piece['title'] );
3665                 $argName = trim( $nameWithSpaces );
3666
3667                 $text = $frame->getArgument( $argName );
3668                 if (  $text === false && ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3669                         # No match in frame, use the supplied default
3670                         $text = $frame->expand( $parts->item( 0 ) );
3671                 }
3672                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3673                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3674                 }
3675
3676                 if ( $text === false ) {
3677                         # No match anywhere
3678                         $text = '{{{' . $frame->implode( '|', $nameWithSpaces, $parts ) . '}}}';
3679                 }
3680                 if ( $error !== false ) {
3681                         $text .= $error;
3682                 }
3683
3684                 wfProfileOut( __METHOD__ );
3685                 return $text;
3686         }
3687
3688         /**
3689          * Return the text to be used for a given extension tag.
3690          * This is the ghost of strip().
3691          *
3692          * @param array $params Associative array of parameters:
3693          *     name       DOMNode for the tag name
3694          *     attr       DOMNode for unparsed text where tag attributes are thought to be
3695          *     attributes Optional associative array of parsed attributes
3696          *     inner      Contents of extension element
3697          *     noClose    Original text did not have a close tag
3698          * @param PPFrame $frame
3699          */
3700         function extensionSubstitution( $params, $frame ) {
3701                 global $wgRawHtml, $wgContLang;
3702                 static $n = 1;
3703
3704                 $name = $frame->expand( $params['name'] );
3705                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3706                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3707
3708                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3709
3710                 if ( $this->ot['html'] ) {
3711                         $name = strtolower( $name );
3712
3713                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3714                         if ( isset( $params['attributes'] ) ) {
3715                                 $attributes = $attributes + $params['attributes'];
3716                         }
3717                         switch ( $name ) {
3718                                 case 'html':
3719                                         if( $wgRawHtml ) {
3720                                                 $output = $content;
3721                                                 break;
3722                                         } else {
3723                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3724                                         }
3725                                 case 'nowiki':
3726                                         $output = Xml::escapeTagsOnly( $content );
3727                                         break;
3728                                 case 'math':
3729                                         $output = $wgContLang->armourMath(
3730                                                 MathRenderer::renderMath( $content, $attributes ) );
3731                                         break;
3732                                 case 'gallery':
3733                                         $output = $this->renderImageGallery( $content, $attributes );
3734                                         break;
3735                                 default:
3736                                         if( isset( $this->mTagHooks[$name] ) ) {
3737                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3738                                                         array( $content, $attributes, $this ) );
3739                                         } else {
3740                                                 throw new MWException( "Invalid call hook $name" );
3741                                         }
3742                         }
3743                 } else {
3744                         if ( $content === null ) {
3745                                 $output = "<$name$attrText/>";
3746                         } else {
3747                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3748                                 $output = "<$name$attrText>$content$close";
3749                         }
3750                 }
3751
3752                 if ( $name == 'html' || $name == 'nowiki' ) {
3753                         $this->mStripState->nowiki->setPair( $marker, $output );
3754                 } else {
3755                         $this->mStripState->general->setPair( $marker, $output );
3756                 }
3757                 return $marker;
3758         }
3759
3760         /**
3761          * Increment an include size counter
3762          *
3763          * @param string $type The type of expansion
3764          * @param integer $size The size of the text
3765          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3766          */
3767         function incrementIncludeSize( $type, $size ) {
3768                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3769                         return false;
3770                 } else {
3771                         $this->mIncludeSizes[$type] += $size;
3772                         return true;
3773                 }
3774         }
3775
3776         /**
3777          * Detect __NOGALLERY__ magic word and set a placeholder
3778          */
3779         function stripNoGallery( &$text ) {
3780                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3781                 # do not add TOC
3782                 $mw = MagicWord::get( 'nogallery' );
3783                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3784         }
3785
3786         /**
3787          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3788          * placeholder that will then be replaced by the real TOC in
3789          * ->formatHeadings, this works because at this points real
3790          * comments will have already been discarded by the sanitizer.
3791          *
3792          * Any additional __TOC__ magic words left over will be discarded
3793          * as there can only be one TOC on the page.
3794          */
3795         function stripToc( $text ) {
3796                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3797                 # do not add TOC
3798                 $mw = MagicWord::get( 'notoc' );
3799                 if( $mw->matchAndRemove( $text ) ) {
3800                         $this->mShowToc = false;
3801                 }
3802
3803                 $mw = MagicWord::get( 'toc' );
3804                 if( $mw->match( $text ) ) {
3805                         $this->mShowToc = true;
3806                         $this->mForceTocPosition = true;
3807
3808                         // Set a placeholder. At the end we'll fill it in with the TOC.
3809                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3810
3811                         // Only keep the first one.
3812                         $text = $mw->replace( '', $text );
3813                 }
3814                 return $text;
3815         }
3816
3817         /**
3818          * This function accomplishes several tasks:
3819          * 1) Auto-number headings if that option is enabled
3820          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3821          * 3) Add a Table of contents on the top for users who have enabled the option
3822          * 4) Auto-anchor headings
3823          *
3824          * It loops through all headlines, collects the necessary data, then splits up the
3825          * string and re-inserts the newly formatted headlines.
3826          *
3827          * @param string $text
3828          * @param boolean $isMain
3829          * @private
3830          */
3831         function formatHeadings( $text, $isMain=true ) {
3832                 global $wgMaxTocLevel, $wgContLang;
3833
3834                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3835                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3836                         $showEditLink = 0;
3837                 } else {
3838                         $showEditLink = $this->mOptions->getEditSection();
3839                 }
3840
3841                 # Inhibit editsection links if requested in the page
3842                 $esw =& MagicWord::get( 'noeditsection' );
3843                 if( $esw->matchAndRemove( $text ) ) {
3844                         $showEditLink = 0;
3845                 }
3846
3847                 # Get all headlines for numbering them and adding funky stuff like [edit]
3848                 # links - this is for later, but we need the number of headlines right now
3849                 $matches = array();
3850                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3851
3852                 # if there are fewer than 4 headlines in the article, do not show TOC
3853                 # unless it's been explicitly enabled.
3854                 $enoughToc = $this->mShowToc &&
3855                         (($numMatches >= 4) || $this->mForceTocPosition);
3856
3857                 # Allow user to stipulate that a page should have a "new section"
3858                 # link added via __NEWSECTIONLINK__
3859                 $mw =& MagicWord::get( 'newsectionlink' );
3860                 if( $mw->matchAndRemove( $text ) )
3861                         $this->mOutput->setNewSection( true );
3862
3863                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3864                 # override above conditions and always show TOC above first header
3865                 $mw =& MagicWord::get( 'forcetoc' );
3866                 if ($mw->matchAndRemove( $text ) ) {
3867                         $this->mShowToc = true;
3868                         $enoughToc = true;
3869                 }
3870
3871                 # We need this to perform operations on the HTML
3872                 $sk = $this->mOptions->getSkin();
3873
3874                 # headline counter
3875                 $headlineCount = 0;
3876                 $numVisible = 0;
3877
3878                 # Ugh .. the TOC should have neat indentation levels which can be
3879                 # passed to the skin functions. These are determined here
3880                 $toc = '';
3881                 $full = '';
3882                 $head = array();
3883                 $sublevelCount = array();
3884                 $levelCount = array();
3885                 $toclevel = 0;
3886                 $level = 0;
3887                 $prevlevel = 0;
3888                 $toclevel = 0;
3889                 $prevtoclevel = 0;
3890                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3891                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3892                 $tocraw = array();
3893
3894                 foreach( $matches[3] as $headline ) {
3895                         $isTemplate = false;
3896                         $titleText = false;
3897                         $sectionIndex = false;
3898                         $numbering = '';
3899                         $markerMatches = array();
3900                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3901                                 $serial = $markerMatches[1];
3902                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3903                                 $isTemplate = ($titleText != $baseTitleText);
3904                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3905                         }
3906
3907                         if( $toclevel ) {
3908                                 $prevlevel = $level;
3909                                 $prevtoclevel = $toclevel;
3910                         }
3911                         $level = $matches[1][$headlineCount];
3912
3913                         if( $doNumberHeadings || $enoughToc ) {
3914
3915                                 if ( $level > $prevlevel ) {
3916                                         # Increase TOC level
3917                                         $toclevel++;
3918                                         $sublevelCount[$toclevel] = 0;
3919                                         if( $toclevel<$wgMaxTocLevel ) {
3920                                                 $prevtoclevel = $toclevel;
3921                                                 $toc .= $sk->tocIndent();
3922                                                 $numVisible++;
3923                                         }
3924                                 }
3925                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3926                                         # Decrease TOC level, find level to jump to
3927
3928                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3929                                                 # Can only go down to level 1
3930                                                 $toclevel = 1;
3931                                         } else {
3932                                                 for ($i = $toclevel; $i > 0; $i--) {
3933                                                         if ( $levelCount[$i] == $level ) {
3934                                                                 # Found last matching level
3935                                                                 $toclevel = $i;
3936                                                                 break;
3937                                                         }
3938                                                         elseif ( $levelCount[$i] < $level ) {
3939                                                                 # Found first matching level below current level
3940                                                                 $toclevel = $i + 1;
3941                                                                 break;
3942                                                         }
3943                                                 }
3944                                         }
3945                                         if( $toclevel<$wgMaxTocLevel ) {
3946                                                 if($prevtoclevel < $wgMaxTocLevel) {
3947                                                         # Unindent only if the previous toc level was shown :p
3948                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3949                                                 } else {
3950                                                         $toc .= $sk->tocLineEnd();
3951                                                 }
3952                                         }
3953                                 }
3954                                 else {
3955                                         # No change in level, end TOC line
3956                                         if( $toclevel<$wgMaxTocLevel ) {
3957                                                 $toc .= $sk->tocLineEnd();
3958                                         }
3959                                 }
3960
3961                                 $levelCount[$toclevel] = $level;
3962
3963                                 # count number of headlines for each level
3964                                 @$sublevelCount[$toclevel]++;
3965                                 $dot = 0;
3966                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3967                                         if( !empty( $sublevelCount[$i] ) ) {
3968                                                 if( $dot ) {
3969                                                         $numbering .= '.';
3970                                                 }
3971                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3972                                                 $dot = 1;
3973                                         }
3974                                 }
3975                         }
3976
3977                         # The safe header is a version of the header text safe to use for links
3978                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3979                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3980
3981                         # Remove link placeholders by the link text.
3982                         #     <!--LINK number-->
3983                         # turns into
3984                         #     link text with suffix
3985                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3986                                                             "\$this->mLinkHolders['texts'][\$1]",
3987                                                             $safeHeadline );
3988                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3989                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3990                                                             $safeHeadline );
3991
3992                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3993                         $tocline = preg_replace(
3994                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3995                                 array( '',                          '<$1>'),
3996                                 $safeHeadline
3997                         );
3998                         $tocline = trim( $tocline );
3999
4000                         # For the anchor, strip out HTML-y stuff period
4001                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4002                         $safeHeadline = trim( $safeHeadline );
4003
4004                         # Save headline for section edit hint before it's escaped
4005                         $headlineHint = $safeHeadline;
4006                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4007                         $refers[$headlineCount] = $safeHeadline;
4008
4009                         # count how many in assoc. array so we can track dupes in anchors
4010                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
4011                         $refcount[$headlineCount] = $refers[$safeHeadline];
4012
4013                         # Don't number the heading if it is the only one (looks silly)
4014                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
4015                                 # the two are different if the line contains a link
4016                                 $headline=$numbering . ' ' . $headline;
4017                         }
4018
4019                         # Create the anchor for linking from the TOC to the section
4020                         $anchor = $safeHeadline;
4021                         if($refcount[$headlineCount] > 1 ) {
4022                                 $anchor .= '_' . $refcount[$headlineCount];
4023                         }
4024                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
4025                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
4026                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
4027                         }
4028                         # give headline the correct <h#> tag
4029                         if( $showEditLink && $sectionIndex !== false ) {
4030                                 if( $isTemplate ) {
4031                                         # Put a T flag in the section identifier, to indicate to extractSections()
4032                                         # that sections inside <includeonly> should be counted.
4033                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
4034                                 } else {
4035                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
4036                                 }
4037                         } else {
4038                                 $editlink = '';
4039                         }
4040                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
4041
4042                         $headlineCount++;
4043                 }
4044
4045                 $this->mOutput->setSections( $tocraw );
4046
4047                 # Never ever show TOC if no headers
4048                 if( $numVisible < 1 ) {
4049                         $enoughToc = false;
4050                 }
4051
4052                 if( $enoughToc ) {
4053                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4054                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
4055                         }
4056                         $toc = $sk->tocList( $toc );
4057                 }
4058
4059                 # split up and insert constructed headlines
4060
4061                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4062                 $i = 0;
4063
4064                 foreach( $blocks as $block ) {
4065                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
4066                                 # This is the [edit] link that appears for the top block of text when
4067                                 # section editing is enabled
4068
4069                                 # Disabled because it broke block formatting
4070                                 # For example, a bullet point in the top line
4071                                 # $full .= $sk->editSectionLink(0);
4072                         }
4073                         $full .= $block;
4074                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4075                                 # Top anchor now in skin
4076                                 $full = $full.$toc;
4077                         }
4078
4079                         if( !empty( $head[$i] ) ) {
4080                                 $full .= $head[$i];
4081                         }
4082                         $i++;
4083                 }
4084                 if( $this->mForceTocPosition ) {
4085                         return str_replace( '<!--MWTOC-->', $toc, $full );
4086                 } else {
4087                         return $full;
4088                 }
4089         }
4090
4091         /**
4092          * Transform wiki markup when saving a page by doing \r\n -> \n
4093          * conversion, substitting signatures, {{subst:}} templates, etc.
4094          *
4095          * @param string $text the text to transform
4096          * @param Title &$title the Title object for the current article
4097          * @param User &$user the User object describing the current user
4098          * @param ParserOptions $options parsing options
4099          * @param bool $clearState whether to clear the parser state first
4100          * @return string the altered wiki markup
4101          * @public
4102          */
4103         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
4104                 $this->mOptions = $options;
4105                 $this->setTitle( $title );
4106                 $this->setOutputType( OT_WIKI );
4107
4108                 if ( $clearState ) {
4109                         $this->clearState();
4110                 }
4111
4112                 $pairs = array(
4113                         "\r\n" => "\n",
4114                 );
4115                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4116                 $text = $this->pstPass2( $text, $user );
4117                 $text = $this->mStripState->unstripBoth( $text );
4118                 return $text;
4119         }
4120
4121         /**
4122          * Pre-save transform helper function
4123          * @private
4124          */
4125         function pstPass2( $text, $user ) {
4126                 global $wgContLang, $wgLocaltimezone;
4127
4128                 /* Note: This is the timestamp saved as hardcoded wikitext to
4129                  * the database, we use $wgContLang here in order to give
4130                  * everyone the same signature and use the default one rather
4131                  * than the one selected in each user's preferences.
4132                  */
4133                 if ( isset( $wgLocaltimezone ) ) {
4134                         $oldtz = getenv( 'TZ' );
4135                         putenv( 'TZ='.$wgLocaltimezone );
4136                 }
4137                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4138                   ' (' . date( 'T' ) . ')';
4139                 if ( isset( $wgLocaltimezone ) ) {
4140                         putenv( 'TZ='.$oldtz );
4141                 }
4142
4143                 # Variable replacement
4144                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4145                 $text = $this->replaceVariables( $text );
4146
4147                 # Strip out <nowiki> etc. added via replaceVariables
4148                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4149
4150                 # Signatures
4151                 $sigText = $this->getUserSig( $user );
4152                 $text = strtr( $text, array(
4153                         '~~~~~' => $d,
4154                         '~~~~' => "$sigText $d",
4155                         '~~~' => $sigText
4156                 ) );
4157
4158                 # Context links: [[|name]] and [[name (context)|]]
4159                 #
4160                 global $wgLegalTitleChars;
4161                 $tc = "[$wgLegalTitleChars]";
4162                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4163
4164                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4165                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4166                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4167
4168                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4169                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4170                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4171
4172                 $t = $this->mTitle->getText();
4173                 $m = array();
4174                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4175                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4176                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4177                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4178                 } else {
4179                         # if there's no context, don't bother duplicating the title
4180                         $text = preg_replace( $p2, '[[\\1]]', $text );
4181                 }
4182
4183                 # Trim trailing whitespace
4184                 $text = rtrim( $text );
4185
4186                 return $text;
4187         }
4188
4189         /**
4190          * Fetch the user's signature text, if any, and normalize to
4191          * validated, ready-to-insert wikitext.
4192          *
4193          * @param User $user
4194          * @return string
4195          * @private
4196          */
4197         function getUserSig( &$user ) {
4198                 global $wgMaxSigChars;
4199
4200                 $username = $user->getName();
4201                 $nickname = $user->getOption( 'nickname' );
4202                 $nickname = $nickname === '' ? $username : $nickname;
4203
4204                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4205                         $nickname = $username;
4206                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4207                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4208                         # Sig. might contain markup; validate this
4209                         if( $this->validateSig( $nickname ) !== false ) {
4210                                 # Validated; clean up (if needed) and return it
4211                                 return $this->cleanSig( $nickname, true );
4212                         } else {
4213                                 # Failed to validate; fall back to the default
4214                                 $nickname = $username;
4215                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4216                         }
4217                 }
4218
4219                 // Make sure nickname doesnt get a sig in a sig
4220                 $nickname = $this->cleanSigInSig( $nickname );
4221
4222                 # If we're still here, make it a link to the user page
4223                 $userText = wfEscapeWikiText( $username );
4224                 $nickText = wfEscapeWikiText( $nickname );
4225                 if ( $user->isAnon() )  {
4226                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4227                 } else {
4228                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4229                 }
4230         }
4231
4232         /**
4233          * Check that the user's signature contains no bad XML
4234          *
4235          * @param string $text
4236          * @return mixed An expanded string, or false if invalid.
4237          */
4238         function validateSig( $text ) {
4239                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4240         }
4241
4242         /**
4243          * Clean up signature text
4244          *
4245          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4246          * 2) Substitute all transclusions
4247          *
4248          * @param string $text
4249          * @param $parsing Whether we're cleaning (preferences save) or parsing
4250          * @return string Signature text
4251          */
4252         function cleanSig( $text, $parsing = false ) {
4253                 if ( !$parsing ) {
4254                         global $wgTitle;
4255                         $this->startExternalParse( $wgTitle, new ParserOptions(), OT_MSG );
4256                 }
4257
4258                 # FIXME: regex doesn't respect extension tags or nowiki
4259                 #  => Move this logic to braceSubstitution()
4260                 $substWord = MagicWord::get( 'subst' );
4261                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4262                 $substText = '{{' . $substWord->getSynonym( 0 );
4263
4264                 $text = preg_replace( $substRegex, $substText, $text );
4265                 $text = $this->cleanSigInSig( $text );
4266                 $dom = $this->preprocessToDom( $text );
4267                 $frame = new PPFrame( $this );
4268                 $text = $frame->expand( $dom->documentElement );
4269
4270                 if ( !$parsing ) {
4271                         $text = $this->mStripState->unstripBoth( $text );
4272                 }
4273
4274                 return $text;
4275         }
4276
4277         /**
4278          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4279          * @param string $text
4280          * @return string Signature text with /~{3,5}/ removed
4281          */
4282         function cleanSigInSig( $text ) {
4283                 $text = preg_replace( '/~{3,5}/', '', $text );
4284                 return $text;
4285         }
4286
4287         /**
4288          * Set up some variables which are usually set up in parse()
4289          * so that an external function can call some class members with confidence
4290          * @public
4291          */
4292         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4293                 $this->setTitle( $title );
4294                 $this->mOptions = $options;
4295                 $this->setOutputType( $outputType );
4296                 if ( $clearState ) {
4297                         $this->clearState();
4298                 }
4299         }
4300
4301         /**
4302          * Transform a MediaWiki message by replacing magic variables.
4303          *
4304          * For some unknown reason, it also expands templates, but only to the
4305          * first recursion level. This is wrong and broken, probably introduced
4306          * accidentally during refactoring, but probably relied upon by thousands
4307          * of users.
4308          *
4309          * @param string $text the text to transform
4310          * @param ParserOptions $options  options
4311          * @return string the text with variables substituted
4312          * @public
4313          */
4314         function transformMsg( $text, $options ) {
4315                 global $wgTitle;
4316                 static $executing = false;
4317
4318                 $fname = "Parser::transformMsg";
4319
4320                 # Guard against infinite recursion
4321                 if ( $executing ) {
4322                         return $text;
4323                 }
4324                 $executing = true;
4325
4326                 wfProfileIn($fname);
4327
4328                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4329                         $this->setTitle( $wgTitle );
4330                 } else {
4331                         $this->setTitle( Title::newFromText('msg') );
4332                 }
4333                 $this->mOptions = $options;
4334                 $this->setOutputType( OT_MSG );
4335                 $this->clearState();
4336                 $text = $this->replaceVariables( $text );
4337                 $text = $this->mStripState->unstripBoth( $text );
4338
4339                 $executing = false;
4340                 wfProfileOut($fname);
4341                 return $text;
4342         }
4343
4344         /**
4345          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4346          * The callback should have the following form:
4347          *    function myParserHook( $text, $params, &$parser ) { ... }
4348          *
4349          * Transform and return $text. Use $parser for any required context, e.g. use
4350          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4351          *
4352          * @public
4353          *
4354          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4355          * @param mixed $callback The callback function (and object) to use for the tag
4356          *
4357          * @return The old value of the mTagHooks array associated with the hook
4358          */
4359         function setHook( $tag, $callback ) {
4360                 $tag = strtolower( $tag );
4361                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4362                 $this->mTagHooks[$tag] = $callback;
4363                 $this->mStripList[] = $tag;
4364
4365                 return $oldVal;
4366         }
4367
4368         function setTransparentTagHook( $tag, $callback ) {
4369                 $tag = strtolower( $tag );
4370                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4371                 $this->mTransparentTagHooks[$tag] = $callback;
4372
4373                 return $oldVal;
4374         }
4375
4376         /**
4377          * Create a function, e.g. {{sum:1|2|3}}
4378          * The callback function should have the form:
4379          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4380          *
4381          * The callback may either return the text result of the function, or an array with the text
4382          * in element 0, and a number of flags in the other elements. The names of the flags are
4383          * specified in the keys. Valid flags are:
4384          *   found                     The text returned is valid, stop processing the template. This
4385          *                             is on by default.
4386          *   nowiki                    Wiki markup in the return value should be escaped
4387          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4388          *
4389          * @public
4390          *
4391          * @param string $id The magic word ID
4392          * @param mixed $callback The callback function (and object) to use
4393          * @param integer $flags a combination of the following flags:
4394          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4395          *
4396          * @return The old callback function for this name, if any
4397          */
4398         function setFunctionHook( $id, $callback, $flags = 0 ) {
4399                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4400                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4401
4402                 # Add to function cache
4403                 $mw = MagicWord::get( $id );
4404                 if( !$mw )
4405                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4406
4407                 $synonyms = $mw->getSynonyms();
4408                 $sensitive = intval( $mw->isCaseSensitive() );
4409
4410                 foreach ( $synonyms as $syn ) {
4411                         # Case
4412                         if ( !$sensitive ) {
4413                                 $syn = strtolower( $syn );
4414                         }
4415                         # Add leading hash
4416                         if ( !( $flags & SFH_NO_HASH ) ) {
4417                                 $syn = '#' . $syn;
4418                         }
4419                         # Remove trailing colon
4420                         if ( substr( $syn, -1, 1 ) == ':' ) {
4421                                 $syn = substr( $syn, 0, -1 );
4422                         }
4423                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4424                 }
4425                 return $oldVal;
4426         }
4427
4428         /**
4429          * Get all registered function hook identifiers
4430          *
4431          * @return array
4432          */
4433         function getFunctionHooks() {
4434                 return array_keys( $this->mFunctionHooks );
4435         }
4436
4437         /**
4438          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4439          * Placeholders created in Skin::makeLinkObj()
4440          * Returns an array of link CSS classes, indexed by PDBK.
4441          * $options is a bit field, RLH_FOR_UPDATE to select for update
4442          */
4443         function replaceLinkHolders( &$text, $options = 0 ) {
4444                 global $wgUser;
4445                 global $wgContLang;
4446
4447                 $fname = 'Parser::replaceLinkHolders';
4448                 wfProfileIn( $fname );
4449
4450                 $pdbks = array();
4451                 $colours = array();
4452                 $linkcolour_ids = array();
4453                 $sk = $this->mOptions->getSkin();
4454                 $linkCache =& LinkCache::singleton();
4455
4456                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4457                         wfProfileIn( $fname.'-check' );
4458                         $dbr = wfGetDB( DB_SLAVE );
4459                         $page = $dbr->tableName( 'page' );
4460                         $threshold = $wgUser->getOption('stubthreshold');
4461
4462                         # Sort by namespace
4463                         asort( $this->mLinkHolders['namespaces'] );
4464
4465                         # Generate query
4466                         $query = false;
4467                         $current = null;
4468                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4469                                 # Make title object
4470                                 $title = $this->mLinkHolders['titles'][$key];
4471
4472                                 # Skip invalid entries.
4473                                 # Result will be ugly, but prevents crash.
4474                                 if ( is_null( $title ) ) {
4475                                         continue;
4476                                 }
4477                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4478
4479                                 # Check if it's a static known link, e.g. interwiki
4480                                 if ( $title->isAlwaysKnown() ) {
4481                                         $colours[$pdbk] = '';
4482                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4483                                         $colours[$pdbk] = '';
4484                                         $this->mOutput->addLink( $title, $id );
4485                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4486                                         $colours[$pdbk] = 'new';
4487                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4488                                         $colours[$pdbk] = 'new';
4489                                 } else {
4490                                         # Not in the link cache, add it to the query
4491                                         if ( !isset( $current ) ) {
4492                                                 $current = $ns;
4493                                                 $query =  "SELECT page_id, page_namespace, page_title";
4494                                                 if ( $threshold > 0 ) {
4495                                                         $query .= ', page_len, page_is_redirect';
4496                                                 }
4497                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4498                                         } elseif ( $current != $ns ) {
4499                                                 $current = $ns;
4500                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4501                                         } else {
4502                                                 $query .= ', ';
4503                                         }
4504
4505                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4506                                 }
4507                         }
4508                         if ( $query ) {
4509                                 $query .= '))';
4510                                 if ( $options & RLH_FOR_UPDATE ) {
4511                                         $query .= ' FOR UPDATE';
4512                                 }
4513
4514                                 $res = $dbr->query( $query, $fname );
4515
4516                                 # Fetch data and form into an associative array
4517                                 # non-existent = broken
4518                                 while ( $s = $dbr->fetchObject($res) ) {
4519                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4520                                         $pdbk = $title->getPrefixedDBkey();
4521                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4522                                         $this->mOutput->addLink( $title, $s->page_id );
4523                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4524                                         //add id to the extension todolist
4525                                         $linkcolour_ids[$s->page_id] = $pdbk;
4526                                 }
4527                                 //pass an array of page_ids to an extension
4528                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4529                         }
4530                         wfProfileOut( $fname.'-check' );
4531
4532                         # Do a second query for different language variants of links and categories
4533                         if($wgContLang->hasVariants()){
4534                                 $linkBatch = new LinkBatch();
4535                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4536                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4537                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4538
4539                                 $categories = $this->mOutput->getCategoryLinks();
4540
4541                                 // Add variants of links to link batch
4542                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4543                                         $title = $this->mLinkHolders['titles'][$key];
4544                                         if ( is_null( $title ) )
4545                                                 continue;
4546
4547                                         $pdbk = $title->getPrefixedDBkey();
4548                                         $titleText = $title->getText();
4549
4550                                         // generate all variants of the link title text
4551                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4552
4553                                         // if link was not found (in first query), add all variants to query
4554                                         if ( !isset($colours[$pdbk]) ){
4555                                                 foreach($allTextVariants as $textVariant){
4556                                                         if($textVariant != $titleText){
4557                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4558                                                                 if(is_null($variantTitle)) continue;
4559                                                                 $linkBatch->addObj( $variantTitle );
4560                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4561                                                         }
4562                                                 }
4563                                         }
4564                                 }
4565
4566                                 // process categories, check if a category exists in some variant
4567                                 foreach( $categories as $category ){
4568                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4569                                         foreach($variants as $variant){
4570                                                 if($variant != $category){
4571                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4572                                                         if(is_null($variantTitle)) continue;
4573                                                         $linkBatch->addObj( $variantTitle );
4574                                                         $categoryMap[$variant] = $category;
4575                                                 }
4576                                         }
4577                                 }
4578
4579
4580                                 if(!$linkBatch->isEmpty()){
4581                                         // construct query
4582                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4583
4584                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4585                                         if ( $threshold > 0 ) {
4586                                                 $variantQuery .= ', page_len, page_is_redirect';
4587                                         }
4588
4589                                         $variantQuery .= " FROM $page WHERE $titleClause";
4590                                         if ( $options & RLH_FOR_UPDATE ) {
4591                                                 $variantQuery .= ' FOR UPDATE';
4592                                         }
4593
4594                                         $varRes = $dbr->query( $variantQuery, $fname );
4595
4596                                         // for each found variants, figure out link holders and replace
4597                                         while ( $s = $dbr->fetchObject($varRes) ) {
4598
4599                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4600                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4601                                                 $vardbk = $variantTitle->getDBkey();
4602
4603                                                 $holderKeys = array();
4604                                                 if(isset($variantMap[$varPdbk])){
4605                                                         $holderKeys = $variantMap[$varPdbk];
4606                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4607                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4608                                                 }
4609
4610                                                 // loop over link holders
4611                                                 foreach($holderKeys as $key){
4612                                                         $title = $this->mLinkHolders['titles'][$key];
4613                                                         if ( is_null( $title ) ) continue;
4614
4615                                                         $pdbk = $title->getPrefixedDBkey();
4616
4617                                                         if(!isset($colours[$pdbk])){
4618                                                                 // found link in some of the variants, replace the link holder data
4619                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4620                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4621
4622                                                                 // set pdbk and colour
4623                                                                 $pdbks[$key] = $varPdbk;
4624                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4625                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4626                                                         }
4627                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4628                                                 }
4629
4630                                                 // check if the object is a variant of a category
4631                                                 if(isset($categoryMap[$vardbk])){
4632                                                         $oldkey = $categoryMap[$vardbk];
4633                                                         if($oldkey != $vardbk)
4634                                                                 $varCategories[$oldkey]=$vardbk;
4635                                                 }
4636                                         }
4637
4638                                         // rebuild the categories in original order (if there are replacements)
4639                                         if(count($varCategories)>0){
4640                                                 $newCats = array();
4641                                                 $originalCats = $this->mOutput->getCategories();
4642                                                 foreach($originalCats as $cat => $sortkey){
4643                                                         // make the replacement
4644                                                         if( array_key_exists($cat,$varCategories) )
4645                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4646                                                         else $newCats[$cat] = $sortkey;
4647                                                 }
4648                                                 $this->mOutput->setCategoryLinks($newCats);
4649                                         }
4650                                 }
4651                         }
4652
4653                         # Construct search and replace arrays
4654                         wfProfileIn( $fname.'-construct' );
4655                         $replacePairs = array();
4656                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4657                                 $pdbk = $pdbks[$key];
4658                                 $searchkey = "<!--LINK $key-->";
4659                                 $title = $this->mLinkHolders['titles'][$key];
4660                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4661                                         $linkCache->addBadLinkObj( $title );
4662                                         $colours[$pdbk] = 'new';
4663                                         $this->mOutput->addLink( $title, 0 );
4664                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4665                                                                         $this->mLinkHolders['texts'][$key],
4666                                                                         $this->mLinkHolders['queries'][$key] );
4667                                 } else {
4668                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4669                                                                         $this->mLinkHolders['texts'][$key],
4670                                                                         $this->mLinkHolders['queries'][$key] );
4671                                 }
4672                         }
4673                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4674                         wfProfileOut( $fname.'-construct' );
4675
4676                         # Do the thing
4677                         wfProfileIn( $fname.'-replace' );
4678                         $text = preg_replace_callback(
4679                                 '/(<!--LINK .*?-->)/',
4680                                 $replacer->cb(),
4681                                 $text);
4682
4683                         wfProfileOut( $fname.'-replace' );
4684                 }
4685
4686                 # Now process interwiki link holders
4687                 # This is quite a bit simpler than internal links
4688                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4689                         wfProfileIn( $fname.'-interwiki' );
4690                         # Make interwiki link HTML
4691                         $replacePairs = array();
4692                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4693                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4694                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4695                         }
4696                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4697
4698                         $text = preg_replace_callback(
4699                                 '/<!--IWLINK (.*?)-->/',
4700                                 $replacer->cb(),
4701                                 $text );
4702                         wfProfileOut( $fname.'-interwiki' );
4703                 }
4704
4705                 wfProfileOut( $fname );
4706                 return $colours;
4707         }
4708
4709         /**
4710          * Replace <!--LINK--> link placeholders with plain text of links
4711          * (not HTML-formatted).
4712          * @param string $text
4713          * @return string
4714          */
4715         function replaceLinkHoldersText( $text ) {
4716                 $fname = 'Parser::replaceLinkHoldersText';
4717                 wfProfileIn( $fname );
4718
4719                 $text = preg_replace_callback(
4720                         '/<!--(LINK|IWLINK) (.*?)-->/',
4721                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4722                         $text );
4723
4724                 wfProfileOut( $fname );
4725                 return $text;
4726         }
4727
4728         /**
4729          * @param array $matches
4730          * @return string
4731          * @private
4732          */
4733         function replaceLinkHoldersTextCallback( $matches ) {
4734                 $type = $matches[1];
4735                 $key  = $matches[2];
4736                 if( $type == 'LINK' ) {
4737                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4738                                 return $this->mLinkHolders['texts'][$key];
4739                         }
4740                 } elseif( $type == 'IWLINK' ) {
4741                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4742                                 return $this->mInterwikiLinkHolders['texts'][$key];
4743                         }
4744                 }
4745                 return $matches[0];
4746         }
4747
4748         /**
4749          * Tag hook handler for 'pre'.
4750          */
4751         function renderPreTag( $text, $attribs ) {
4752                 // Backwards-compatibility hack
4753                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4754
4755                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4756                 return wfOpenElement( 'pre', $attribs ) .
4757                         Xml::escapeTagsOnly( $content ) .
4758                         '</pre>';
4759         }
4760
4761         /**
4762          * Renders an image gallery from a text with one line per image.
4763          * text labels may be given by using |-style alternative text. E.g.
4764          *   Image:one.jpg|The number "1"
4765          *   Image:tree.jpg|A tree
4766          * given as text will return the HTML of a gallery with two images,
4767          * labeled 'The number "1"' and
4768          * 'A tree'.
4769          */
4770         function renderImageGallery( $text, $params ) {
4771                 $ig = new ImageGallery();
4772                 $ig->setContextTitle( $this->mTitle );
4773                 $ig->setShowBytes( false );
4774                 $ig->setShowFilename( false );
4775                 $ig->setParser( $this );
4776                 $ig->setHideBadImages();
4777                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4778                 $ig->useSkin( $this->mOptions->getSkin() );
4779                 $ig->mRevisionId = $this->mRevisionId;
4780
4781                 if( isset( $params['caption'] ) ) {
4782                         $caption = $params['caption'];
4783                         $caption = htmlspecialchars( $caption );
4784                         $caption = $this->replaceInternalLinks( $caption );
4785                         $ig->setCaptionHtml( $caption );
4786                 }
4787                 if( isset( $params['perrow'] ) ) {
4788                         $ig->setPerRow( $params['perrow'] );
4789                 }
4790                 if( isset( $params['widths'] ) ) {
4791                         $ig->setWidths( $params['widths'] );
4792                 }
4793                 if( isset( $params['heights'] ) ) {
4794                         $ig->setHeights( $params['heights'] );
4795                 }
4796
4797                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4798
4799                 $lines = explode( "\n", $text );
4800                 foreach ( $lines as $line ) {
4801                         # match lines like these:
4802                         # Image:someimage.jpg|This is some image
4803                         $matches = array();
4804                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4805                         # Skip empty lines
4806                         if ( count( $matches ) == 0 ) {
4807                                 continue;
4808                         }
4809                         $tp = Title::newFromText( $matches[1] );
4810                         $nt =& $tp;
4811                         if( is_null( $nt ) ) {
4812                                 # Bogus title. Ignore these so we don't bomb out later.
4813                                 continue;
4814                         }
4815                         if ( isset( $matches[3] ) ) {
4816                                 $label = $matches[3];
4817                         } else {
4818                                 $label = '';
4819                         }
4820
4821                         $html = $this->recursiveTagParse( trim( $label ) );
4822
4823                         $ig->add( $nt, $html );
4824
4825                         # Only add real images (bug #5586)
4826                         if ( $nt->getNamespace() == NS_IMAGE ) {
4827                                 $this->mOutput->addImage( $nt->getDBkey() );
4828                         }
4829                 }
4830                 return $ig->toHTML();
4831         }
4832
4833         function getImageParams( $handler ) {
4834                 if ( $handler ) {
4835                         $handlerClass = get_class( $handler );
4836                 } else {
4837                         $handlerClass = '';
4838                 }
4839                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4840                         // Initialise static lists
4841                         static $internalParamNames = array(
4842                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4843                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4844                                         'bottom', 'text-bottom' ),
4845                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4846                                         'upright', 'border' ),
4847                         );
4848                         static $internalParamMap;
4849                         if ( !$internalParamMap ) {
4850                                 $internalParamMap = array();
4851                                 foreach ( $internalParamNames as $type => $names ) {
4852                                         foreach ( $names as $name ) {
4853                                                 $magicName = str_replace( '-', '_', "img_$name" );
4854                                                 $internalParamMap[$magicName] = array( $type, $name );
4855                                         }
4856                                 }
4857                         }
4858
4859                         // Add handler params
4860                         $paramMap = $internalParamMap;
4861                         if ( $handler ) {
4862                                 $handlerParamMap = $handler->getParamMap();
4863                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4864                                         $paramMap[$magic] = array( 'handler', $paramName );
4865                                 }
4866                         }
4867                         $this->mImageParams[$handlerClass] = $paramMap;
4868                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4869                 }
4870                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4871         }
4872
4873         /**
4874          * Parse image options text and use it to make an image
4875          */
4876         function makeImage( $title, $options ) {
4877                 # @TODO: let the MediaHandler specify its transform parameters
4878                 #
4879                 # Check if the options text is of the form "options|alt text"
4880                 # Options are:
4881                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4882                 #  * left               no resizing, just left align. label is used for alt= only
4883                 #  * right              same, but right aligned
4884                 #  * none               same, but not aligned
4885                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4886                 #  * center             center the image
4887                 #  * framed             Keep original image size, no magnify-button.
4888                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4889                 #  * upright            reduce width for upright images, rounded to full __0 px
4890                 #  * border             draw a 1px border around the image
4891                 # vertical-align values (no % or length right now):
4892                 #  * baseline
4893                 #  * sub
4894                 #  * super
4895                 #  * top
4896                 #  * text-top
4897                 #  * middle
4898                 #  * bottom
4899                 #  * text-bottom
4900
4901                 $parts = array_map( 'trim', explode( '|', $options) );
4902                 $sk = $this->mOptions->getSkin();
4903
4904                 # Give extensions a chance to select the file revision for us
4905                 $skip = $time = false;
4906                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4907
4908                 if ( $skip ) {
4909                         return $sk->makeLinkObj( $title );
4910                 }
4911
4912                 # Get parameter map
4913                 $file = wfFindFile( $title, $time );
4914                 $handler = $file ? $file->getHandler() : false;
4915
4916                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4917
4918                 # Process the input parameters
4919                 $caption = '';
4920                 $params = array( 'frame' => array(), 'handler' => array(),
4921                         'horizAlign' => array(), 'vertAlign' => array() );
4922                 foreach( $parts as $part ) {
4923                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4924                         if ( isset( $paramMap[$magicName] ) ) {
4925                                 list( $type, $paramName ) = $paramMap[$magicName];
4926                                 $params[$type][$paramName] = $value;
4927
4928                                 // Special case; width and height come in one variable together
4929                                 if( $type == 'handler' && $paramName == 'width' ) {
4930                                         $m = array();
4931                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4932                                                 $params[$type]['width'] = intval( $m[1] );
4933                                                 $params[$type]['height'] = intval( $m[2] );
4934                                         } else {
4935                                                 $params[$type]['width'] = intval( $value );
4936                                         }
4937                                 }
4938                         } else {
4939                                 $caption = $part;
4940                         }
4941                 }
4942
4943                 # Process alignment parameters
4944                 if ( $params['horizAlign'] ) {
4945                         $params['frame']['align'] = key( $params['horizAlign'] );
4946                 }
4947                 if ( $params['vertAlign'] ) {
4948                         $params['frame']['valign'] = key( $params['vertAlign'] );
4949                 }
4950
4951                 # Validate the handler parameters
4952                 if ( $handler ) {
4953                         foreach ( $params['handler'] as $name => $value ) {
4954                                 if ( !$handler->validateParam( $name, $value ) ) {
4955                                         unset( $params['handler'][$name] );
4956                                 }
4957                         }
4958                 }
4959
4960                 # Strip bad stuff out of the alt text
4961                 $alt = $this->replaceLinkHoldersText( $caption );
4962
4963                 # make sure there are no placeholders in thumbnail attributes
4964                 # that are later expanded to html- so expand them now and
4965                 # remove the tags
4966                 $alt = $this->mStripState->unstripBoth( $alt );
4967                 $alt = Sanitizer::stripAllTags( $alt );
4968
4969                 $params['frame']['alt'] = $alt;
4970                 $params['frame']['caption'] = $caption;
4971
4972                 # Linker does the rest
4973                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4974
4975                 # Give the handler a chance to modify the parser object
4976                 if ( $handler ) {
4977                         $handler->parserTransformHook( $this, $file );
4978                 }
4979
4980                 return $ret;
4981         }
4982
4983         /**
4984          * Set a flag in the output object indicating that the content is dynamic and
4985          * shouldn't be cached.
4986          */
4987         function disableCache() {
4988                 wfDebug( "Parser output marked as uncacheable.\n" );
4989                 $this->mOutput->mCacheTime = -1;
4990         }
4991
4992         /**#@+
4993          * Callback from the Sanitizer for expanding items found in HTML attribute
4994          * values, so they can be safely tested and escaped.
4995          * @param string $text
4996          * @param PPFrame $frame
4997          * @return string
4998          * @private
4999          */
5000         function attributeStripCallback( &$text, $frame = false ) {
5001                 $text = $this->replaceVariables( $text, $frame );
5002                 $text = $this->mStripState->unstripBoth( $text );
5003                 return $text;
5004         }
5005
5006         /**#@-*/
5007
5008         /**#@+
5009          * Accessor/mutator
5010          */
5011         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
5012         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
5013         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
5014         /**#@-*/
5015
5016         /**#@+
5017          * Accessor
5018          */
5019         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
5020         /**#@-*/
5021
5022
5023         /**
5024          * Break wikitext input into sections, and either pull or replace
5025          * some particular section's text.
5026          *
5027          * External callers should use the getSection and replaceSection methods.
5028          *
5029          * @param string $text Page wikitext
5030          * @param string $section A section identifier string of the form:
5031          *   <flag1> - <flag2> - ... - <section number>
5032          *
5033          * Currently the only recognised flag is "T", which means the target section number
5034          * was derived during a template inclusion parse, in other words this is a template
5035          * section edit link. If no flags are given, it was an ordinary section edit link.
5036          * This flag is required to avoid a section numbering mismatch when a section is
5037          * enclosed by <includeonly> (bug 6563).
5038          *
5039          * The section number 0 pulls the text before the first heading; other numbers will
5040          * pull the given section along with its lower-level subsections. If the section is
5041          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5042          *
5043          * @param string $mode One of "get" or "replace"
5044          * @param string $newText Replacement text for section data.
5045          * @return string for "get", the extracted section text.
5046          *                for "replace", the whole page with the section replaced.
5047          */
5048         private function extractSections( $text, $section, $mode, $newText='' ) {
5049                 global $wgTitle;
5050                 $this->clearState();
5051                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
5052                 $this->mOptions = new ParserOptions;
5053                 $this->setOutputType( OT_WIKI );
5054                 $curIndex = 0;
5055                 $outText = '';
5056                 $frame = new PPFrame( $this );
5057
5058                 // Process section extraction flags
5059                 $flags = 0;
5060                 $sectionParts = explode( '-', $section );
5061                 $sectionIndex = array_pop( $sectionParts );
5062                 foreach ( $sectionParts as $part ) {
5063                         if ( $part == 'T' ) {
5064                                 $flags |= self::PTD_FOR_INCLUSION;
5065                         }
5066                 }
5067                 // Preprocess the text
5068                 $dom = $this->preprocessToDom( $text, $flags );
5069                 $root = $dom->documentElement;
5070
5071                 // <h> nodes indicate section breaks
5072                 // They can only occur at the top level, so we can find them by iterating the root's children
5073                 $node = $root->firstChild;
5074
5075                 // Find the target section
5076                 if ( $sectionIndex == 0 ) {
5077                         // Section zero doesn't nest, level=big
5078                         $targetLevel = 1000;
5079                 } else {
5080                         while ( $node ) {
5081                                 if ( $node->nodeName == 'h' ) {
5082                                         if ( $curIndex + 1 == $sectionIndex ) {
5083                                                 break;
5084                                         }
5085                                         $curIndex++;
5086                                 }
5087                                 if ( $mode == 'replace' ) {
5088                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5089                                 }
5090                                 $node = $node->nextSibling;
5091                         }
5092                         if ( $node ) {
5093                                 $targetLevel = $node->getAttribute( 'level' );
5094                         }
5095                 }
5096
5097                 if ( !$node ) {
5098                         // Not found
5099                         if ( $mode == 'get' ) {
5100                                 return $newText;
5101                         } else {
5102                                 return $text;
5103                         }
5104                 }
5105
5106                 // Find the end of the section, including nested sections
5107                 do {
5108                         if ( $node->nodeName == 'h' ) {
5109                                 $curIndex++;
5110                                 $curLevel = $node->getAttribute( 'level' );
5111                                 if ( $curIndex != $sectionIndex && $curLevel <= $targetLevel ) {
5112                                         break;
5113                                 }
5114                         }
5115                         if ( $mode == 'get' ) {
5116                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5117                         }
5118                         $node = $node->nextSibling;
5119                 } while ( $node );
5120
5121                 // Write out the remainder (in replace mode only)
5122                 if ( $mode == 'replace' ) {
5123                         // Output the replacement text
5124                         // Add two newlines on -- trailing whitespace in $newText is conventionally
5125                         // stripped by the editor, so we need both newlines to restore the paragraph gap
5126                         $outText .= $newText . "\n\n";
5127                         while ( $node ) {
5128                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5129                                 $node = $node->nextSibling;
5130                         }
5131                 }
5132
5133                 if ( is_string( $outText ) ) {
5134                         // Re-insert stripped tags
5135                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5136                 }
5137
5138                 return $outText;
5139         }
5140
5141         /**
5142          * This function returns the text of a section, specified by a number ($section).
5143          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5144          * the first section before any such heading (section 0).
5145          *
5146          * If a section contains subsections, these are also returned.
5147          *
5148          * @param string $text text to look in
5149          * @param string $section section identifier
5150          * @param string $deftext default to return if section is not found
5151          * @return string text of the requested section
5152          */
5153         public function getSection( $text, $section, $deftext='' ) {
5154                 return $this->extractSections( $text, $section, "get", $deftext );
5155         }
5156
5157         public function replaceSection( $oldtext, $section, $text ) {
5158                 return $this->extractSections( $oldtext, $section, "replace", $text );
5159         }
5160
5161         /**
5162          * Get the timestamp associated with the current revision, adjusted for
5163          * the default server-local timestamp
5164          */
5165         function getRevisionTimestamp() {
5166                 if ( is_null( $this->mRevisionTimestamp ) ) {
5167                         wfProfileIn( __METHOD__ );
5168                         global $wgContLang;
5169                         $dbr = wfGetDB( DB_SLAVE );
5170                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5171                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5172
5173                         // Normalize timestamp to internal MW format for timezone processing.
5174                         // This has the added side-effect of replacing a null value with
5175                         // the current time, which gives us more sensible behavior for
5176                         // previews.
5177                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5178
5179                         // The cryptic '' timezone parameter tells to use the site-default
5180                         // timezone offset instead of the user settings.
5181                         //
5182                         // Since this value will be saved into the parser cache, served
5183                         // to other users, and potentially even used inside links and such,
5184                         // it needs to be consistent for all visitors.
5185                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5186
5187                         wfProfileOut( __METHOD__ );
5188                 }
5189                 return $this->mRevisionTimestamp;
5190         }
5191
5192         /**
5193          * Mutator for $mDefaultSort
5194          *
5195          * @param $sort New value
5196          */
5197         public function setDefaultSort( $sort ) {
5198                 $this->mDefaultSort = $sort;
5199         }
5200
5201         /**
5202          * Accessor for $mDefaultSort
5203          * Will use the title/prefixed title if none is set
5204          *
5205          * @return string
5206          */
5207         public function getDefaultSort() {
5208                 if( $this->mDefaultSort !== false ) {
5209                         return $this->mDefaultSort;
5210                 } else {
5211                         return $this->mTitle->getNamespace() == NS_CATEGORY
5212                                         ? $this->mTitle->getText()
5213                                         : $this->mTitle->getPrefixedText();
5214                 }
5215         }
5216
5217         /**
5218          * Try to guess the section anchor name based on a wikitext fragment
5219          * presumably extracted from a heading, for example "Header" from
5220          * "== Header ==".
5221          */
5222         public function guessSectionNameFromWikiText( $text ) {
5223                 # Strip out wikitext links(they break the anchor)
5224                 $text = $this->stripSectionName( $text );
5225                 $headline = Sanitizer::decodeCharReferences( $text );
5226                 # strip out HTML
5227                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5228                 $headline = trim( $headline );
5229                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5230                 $replacearray = array(
5231                         '%3A' => ':',
5232                         '%' => '.'
5233                 );
5234                 return str_replace(
5235                         array_keys( $replacearray ),
5236                         array_values( $replacearray ),
5237                         $sectionanchor );
5238         }
5239
5240         /**
5241          * Strips a text string of wikitext for use in a section anchor
5242          *
5243          * Accepts a text string and then removes all wikitext from the
5244          * string and leaves only the resultant text (i.e. the result of
5245          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5246          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5247          * to create valid section anchors by mimicing the output of the
5248          * parser when headings are parsed.
5249          *
5250          * @param $text string Text string to be stripped of wikitext
5251          * for use in a Section anchor
5252          * @return Filtered text string
5253          */
5254         public function stripSectionName( $text ) {
5255                 # Strip internal link markup
5256                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5257                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5258
5259                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5260                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5261                 # on how many empty links there are on the page - need to figure that out.
5262                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5263
5264                 # Parse wikitext quotes (italics & bold)
5265                 $text = $this->doQuotes($text);
5266
5267                 # Strip HTML tags
5268                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5269                 return $text;
5270         }
5271
5272         /**
5273          * strip/replaceVariables/unstrip for preprocessor regression testing
5274          */
5275         function srvus( $text ) {
5276                 $text = $this->replaceVariables( $text );
5277                 $text = $this->mStripState->unstripBoth( $text );
5278                 $text = Sanitizer::removeHTMLtags( $text );
5279                 return $text;
5280         }
5281 }
5282
5283 /**
5284  * @todo document, briefly.
5285  * @addtogroup Parser
5286  */
5287 class OnlyIncludeReplacer {
5288         var $output = '';
5289
5290         function replace( $matches ) {
5291                 if ( substr( $matches[1], -1 ) == "\n" ) {
5292                         $this->output .= substr( $matches[1], 0, -1 );
5293                 } else {
5294                         $this->output .= $matches[1];
5295                 }
5296         }
5297 }
5298
5299 /**
5300  * @todo document, briefly.
5301  * @addtogroup Parser
5302  */
5303 class StripState {
5304         var $general, $nowiki;
5305
5306         function __construct() {
5307                 $this->general = new ReplacementArray;
5308                 $this->nowiki = new ReplacementArray;
5309         }
5310
5311         function unstripGeneral( $text ) {
5312                 wfProfileIn( __METHOD__ );
5313                 do {
5314                         $oldText = $text;
5315                         $text = $this->general->replace( $text );
5316                 } while ( $text != $oldText );
5317                 wfProfileOut( __METHOD__ );
5318                 return $text;
5319         }
5320
5321         function unstripNoWiki( $text ) {
5322                 wfProfileIn( __METHOD__ );
5323                 do {
5324                         $oldText = $text;
5325                         $text = $this->nowiki->replace( $text );
5326                 } while ( $text != $oldText );
5327                 wfProfileOut( __METHOD__ );
5328                 return $text;
5329         }
5330
5331         function unstripBoth( $text ) {
5332                 wfProfileIn( __METHOD__ );
5333                 do {
5334                         $oldText = $text;
5335                         $text = $this->general->replace( $text );
5336                         $text = $this->nowiki->replace( $text );
5337                 } while ( $text != $oldText );
5338                 wfProfileOut( __METHOD__ );
5339                 return $text;
5340         }
5341 }
5342
5343 /**
5344  * An expansion frame, used as a context to expand the result of preprocessToDom()
5345  */
5346 class PPFrame {
5347         var $parser, $title;
5348         var $titleCache;
5349
5350         /**
5351          * Hashtable listing templates which are disallowed for expansion in this frame,
5352          * having been encountered previously in parent frames.
5353          */
5354         var $loopCheckHash;
5355
5356         /**
5357          * Recursion depth of this frame, top = 0
5358          */
5359         var $depth;
5360
5361         const NO_ARGS = 1;
5362         const NO_TEMPLATES = 2;
5363         const STRIP_COMMENTS = 4;
5364         const NO_IGNORE = 8;
5365
5366         const RECOVER_ORIG = 11;
5367
5368         /**
5369          * Construct a new preprocessor frame.
5370          * @param Parser $parser The parent parser
5371          * @param Title $title The context title, or false if there isn't one
5372          */
5373         function __construct( $parser ) {
5374                 $this->parser = $parser;
5375                 $this->title = $parser->mTitle;
5376                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
5377                 $this->loopCheckHash = array();
5378                 $this->depth = 0;
5379         }
5380
5381         /**
5382          * Create a new child frame
5383          * $args is optionally a DOMNodeList containing the template arguments
5384          */
5385         function newChild( $args = false, $title = false ) {
5386                 $namedArgs = array();
5387                 $numberedArgs = array();
5388                 if ( $title === false ) {
5389                         $title = $this->title;
5390                 }
5391                 if ( $args !== false ) {
5392                         $xpath = false;
5393                         foreach ( $args as $arg ) {
5394                                 if ( !$xpath ) {
5395                                         $xpath = new DOMXPath( $arg->ownerDocument );
5396                                 }
5397
5398                                 $nameNodes = $xpath->query( 'name', $arg );
5399                                 $value = $xpath->query( 'value', $arg );
5400                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5401                                         // Numbered parameter
5402                                         $index = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5403                                         $numberedArgs[$index] = $value->item( 0 );
5404                                         unset( $namedArgs[$index] );
5405                                 } else {
5406                                         // Named parameter
5407                                         $name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS ) );
5408                                         $namedArgs[$name] = $value->item( 0 );
5409                                         unset( $numberedArgs[$name] );
5410                                 }
5411                         }
5412                 }
5413                 return new PPTemplateFrame( $this->parser, $this, $numberedArgs, $namedArgs, $title );
5414         }
5415
5416         /**
5417          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5418          * using the current context
5419          * @param $root the node
5420          */
5421         function expand( $root, $flags = 0 ) {
5422                 if ( is_string( $root ) ) {
5423                         return $root;
5424                 }
5425
5426                 if ( $this->parser->ot['html']
5427                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5428                 {
5429                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5430                 }
5431
5432                 if ( is_array( $root ) ) {
5433                         $s = '';
5434                         foreach ( $root as $node ) {
5435                                 $s .= $this->expand( $node, $flags );
5436                         }
5437                 } elseif ( $root instanceof DOMNodeList ) {
5438                         $s = '';
5439                         foreach ( $root as $node ) {
5440                                 $s .= $this->expand( $node, $flags );
5441                         }
5442                 } elseif ( $root instanceof DOMNode ) {
5443                         if ( $root->nodeType == XML_TEXT_NODE ) {
5444                                 $s = $root->nodeValue;
5445                         } elseif ( $root->nodeName == 'template' ) {
5446                                 # Double-brace expansion
5447                                 $xpath = new DOMXPath( $root->ownerDocument );
5448                                 $titles = $xpath->query( 'title', $root );
5449                                 $title = $titles->item( 0 );
5450                                 $parts = $xpath->query( 'part', $root );
5451                                 if ( $flags & self::NO_TEMPLATES ) {
5452                                         $s = '{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}';
5453                                 } else {
5454                                         $lineStart = $root->getAttribute( 'lineStart' );
5455                                         $params = array(
5456                                                 'title' => $title,
5457                                                 'parts' => $parts,
5458                                                 'lineStart' => $lineStart,
5459                                                 'text' => 'FIXME' );
5460                                         $s = $this->parser->braceSubstitution( $params, $this );
5461                                 }
5462                         } elseif ( $root->nodeName == 'tplarg' ) {
5463                                 # Triple-brace expansion
5464                                 $xpath = new DOMXPath( $root->ownerDocument );
5465                                 $titles = $xpath->query( 'title', $root );
5466                                 $title = $titles->item( 0 );
5467                                 $parts = $xpath->query( 'part', $root );
5468                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5469                                         $s = '{{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}}';
5470                                 } else {
5471                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5472                                         $s = $this->parser->argSubstitution( $params, $this );
5473                                 }
5474                         } elseif ( $root->nodeName == 'comment' ) {
5475                                 # HTML-style comment
5476                                 if ( $this->parser->ot['html']
5477                                         || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
5478                                         || ( $flags & self::STRIP_COMMENTS ) )
5479                                 {
5480                                         $s = '';
5481                                 } else {
5482                                         $s = $root->textContent;
5483                                 }
5484                         } elseif ( $root->nodeName == 'ignore' ) {
5485                                 # Output suppression used by <includeonly> etc.
5486                                 # OT_WIKI will only respect <ignore> in substed templates.
5487                                 # The other output types respect it unless NO_IGNORE is set.
5488                                 # extractSections() sets NO_IGNORE and so never respects it.
5489                                 if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & self::NO_IGNORE ) ) {
5490                                         $s = $root->textContent;
5491                                 } else {
5492                                         $s = '';
5493                                 }
5494                         } elseif ( $root->nodeName == 'ext' ) {
5495                                 # Extension tag
5496                                 $xpath = new DOMXPath( $root->ownerDocument );
5497                                 $names = $xpath->query( 'name', $root );
5498                                 $attrs = $xpath->query( 'attr', $root );
5499                                 $inners = $xpath->query( 'inner', $root );
5500                                 $closes = $xpath->query( 'close', $root );
5501                                 $params = array(
5502                                         'name' => $names->item( 0 ),
5503                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5504                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5505                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5506                                 );
5507                                 $s = $this->parser->extensionSubstitution( $params, $this );
5508                         } elseif ( $root->nodeName == 'h' ) {
5509                                 # Heading
5510                                 $s = $this->expand( $root->childNodes, $flags );
5511
5512                                 if ( $this->parser->ot['html'] ) {
5513                                         # Insert heading index marker
5514                                         $headingIndex = $root->getAttribute( 'i' );
5515                                         $titleText = $this->title->getPrefixedDBkey();
5516                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5517                                         $serial = count( $this->parser->mHeadings ) - 1;
5518                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5519                                         $count = $root->getAttribute( 'level' );
5520                                         $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5521                                         $this->parser->mStripState->general->setPair( $marker, '' );
5522                                 }
5523                         } else {
5524                                 # Generic recursive expansion
5525                                 $s = '';
5526                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5527                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5528                                                 $s .= $node->nodeValue;
5529                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5530                                                 $s .= $this->expand( $node, $flags );
5531                                         }
5532                                 }
5533                         }
5534                 } else {
5535                         throw new MWException( __METHOD__.': Invalid parameter type' );
5536                 }
5537                 return $s;
5538         }
5539
5540         function implodeWithFlags( $sep, $flags /*, ... */ ) {
5541                 $args = array_slice( func_get_args(), 2 );
5542
5543                 $first = true;
5544                 $s = '';
5545                 foreach ( $args as $root ) {
5546                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5547                                 $root = array( $root );
5548                         }
5549                         foreach ( $root as $node ) {
5550                                 if ( $first ) {
5551                                         $first = false;
5552                                 } else {
5553                                         $s .= $sep;
5554                                 }
5555                                 $s .= $this->expand( $node, $flags );
5556                         }
5557                 }
5558                 return $s;
5559         }
5560
5561         function implode( $sep /*, ... */ ) {
5562                 $args = func_get_args();
5563                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0 ), array_slice( $args, 1 ) );
5564                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5565         }
5566
5567         /**
5568          * Split an <arg> or <template> node into a three-element array:
5569          *    DOMNode name, string index and DOMNode value
5570          */
5571         function splitBraceNode( $node ) {
5572                 $xpath = new DOMXPath( $node->ownerDocument );
5573                 $names = $xpath->query( 'name', $node );
5574                 $values = $xpath->query( 'value', $node );
5575                 if ( !$names->length || !$values->length ) {
5576                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
5577                 }
5578                 $name = $names->item( 0 );
5579                 $index = $name->getAttribute( 'index' );
5580                 return array( $name, $index, $values->item( 0 ) );
5581         }
5582
5583         /**
5584          * Split an <ext> node into an associative array containing name, attr, inner and close
5585          * All values in the resulting array are DOMNodes. Inner and close are optional.
5586          */
5587         function splitExtNode( $node ) {
5588                 $xpath = new DOMXPath( $node->ownerDocument );
5589                 $names = $xpath->query( 'name', $node );
5590                 $attrs = $xpath->query( 'attr', $node );
5591                 $inners = $xpath->query( 'inner', $node );
5592                 $closes = $xpath->query( 'close', $node );
5593                 if ( !$names->length || !$attrs->length ) {
5594                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
5595                 }
5596                 $parts = array(
5597                         'name' => $names->item( 0 ),
5598                         'attr' => $attrs->item( 0 ) );
5599                 if ( $inners->length ) {
5600                         $parts['inner'] = $inners->item( 0 );
5601                 }
5602                 if ( $closes->length ) {
5603                         $parts['close'] = $closes->item( 0 );
5604                 }
5605                 return $parts;
5606         }
5607
5608         function __toString() {
5609                 return 'frame{}';
5610         }
5611
5612         function getPDBK( $level = false ) {
5613                 if ( $level === false ) {
5614                         return $this->title->getPrefixedDBkey();
5615                 } else {
5616                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
5617                 }
5618         }
5619
5620         /**
5621          * Returns true if there are no arguments in this frame
5622          */
5623         function isEmpty() {
5624                 return true;
5625         }
5626
5627         function getArgument( $name ) {
5628                 return false;
5629         }
5630
5631         /**
5632          * Returns true if the infinite loop check is OK, false if a loop is detected
5633          */
5634         function loopCheck( $title ) {
5635                 return !isset( $this->loopCheckHash[$title->getPrefixedDBkey()] );
5636         }
5637 }
5638
5639 /**
5640  * Expansion frame with template arguments
5641  */
5642 class PPTemplateFrame extends PPFrame {
5643         var $numberedArgs, $namedArgs, $parent;
5644         var $numberedExpansionCache, $namedExpansionCache;
5645
5646         function __construct( $parser, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
5647                 $this->parser = $parser;
5648                 $this->parent = $parent;
5649                 $this->numberedArgs = $numberedArgs;
5650                 $this->namedArgs = $namedArgs;
5651                 $this->title = $title;
5652                 $pdbk = $title ? $title->getPrefixedDBkey() : false;
5653                 $this->titleCache = $parent->titleCache;
5654                 $this->titleCache[] = $pdbk;
5655                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
5656                 if ( $pdbk !== false ) {
5657                         $this->loopCheckHash[$pdbk] = true;
5658                 }
5659                 $this->depth = $parent->depth + 1;
5660                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
5661         }
5662
5663         function __toString() {
5664                 $s = 'tplframe{';
5665                 $first = true;
5666                 $args = $this->numberedArgs + $this->namedArgs;
5667                 foreach ( $args as $name => $value ) {
5668                         if ( $first ) {
5669                                 $first = false;
5670                         } else {
5671                                 $s .= ', ';
5672                         }
5673                         $s .= "\"$name\":\"" .
5674                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5675                 }
5676                 $s .= '}';
5677                 return $s;
5678         }
5679         /**
5680          * Returns true if there are no arguments in this frame
5681          */
5682         function isEmpty() {
5683                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
5684         }
5685
5686         function getNumberedArgument( $index ) {
5687                 if ( !isset( $this->numberedArgs[$index] ) ) {
5688                         return false;
5689                 }
5690                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
5691                         # No trimming for unnamed arguments
5692                         $this->numberedExpansionCache[$index] = $this->parent->expand( $this->numberedArgs[$index], self::STRIP_COMMENTS );
5693                 }
5694                 return $this->numberedExpansionCache[$index];
5695         }
5696
5697         function getNamedArgument( $name ) {
5698                 if ( !isset( $this->namedArgs[$name] ) ) {
5699                         return false;
5700                 }
5701                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
5702                         # Trim named arguments post-expand, for backwards compatibility
5703                         $this->namedExpansionCache[$name] = trim(
5704                                 $this->parent->expand( $this->namedArgs[$name], self::STRIP_COMMENTS ) );
5705                 }
5706                 return $this->namedExpansionCache[$name];
5707         }
5708
5709         function getArgument( $name ) {
5710                 wfDebug( __METHOD__." getting '$name'\n" );
5711                 $text = $this->getNumberedArgument( $name );
5712                 if ( $text === false ) {
5713                         $text = $this->getNamedArgument( $name );
5714                 }
5715                 return $text;
5716         }
5717 }