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          * Accessor for mUniqPrefix.
 255          *
 256          * @public
 257          */
 258         function uniqPrefix() {
 259                 if( !isset( $this->mUniqPrefix ) ) {
 260                         // @fixme this is probably *horribly wrong*
 261                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 262                         // if this is called for a parser cache hit, the parser may not
 263                         // have ever been initialized in the first place.
 264                         // Not really sure what the heck is supposed to be going on here.
 265                         return '';
 266                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 267                 }
 268                 return $this->mUniqPrefix;
 269         }
 270
 271         /**
 272          * Convert wikitext to HTML
 273          * Do not call this function recursively.
 274          *
 275          * @param string $text Text we want to parse
 276          * @param Title &$title A title object
 277          * @param array $options
 278          * @param boolean $linestart
 279          * @param boolean $clearState
 280          * @param int $revid number to pass in {{REVISIONID}}
 281          * @return ParserOutput a ParserOutput
 282          */
 283         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 284                 /**
 285                  * First pass--just handle <nowiki> sections, pass the rest off
 286                  * to internalParse() which does all the real work.
 287                  */
 288
 289                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 290                 $fname = 'Parser::parse-' . wfGetCaller();
 291                 wfProfileIn( __METHOD__ );
 292                 wfProfileIn( $fname );
 293
 294                 if ( $clearState ) {
 295                         $this->clearState();
 296                 }
 297
 298                 $this->mOptions = $options;
 299                 $this->mTitle =& $title;
 300                 $oldRevisionId = $this->mRevisionId;
 301                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 302                 if( $revid !== null ) {
 303                         $this->mRevisionId = $revid;
 304                         $this->mRevisionTimestamp = null;
 305                 }
 306                 $this->setOutputType( OT_HTML );
 307                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 308                 # No more strip!
 309                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 310                 $text = $this->internalParse( $text );
 311                 $text = $this->mStripState->unstripGeneral( $text );
 312
 313                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 314                 $fixtags = array(
 315                         # french spaces, last one Guillemet-left
 316                         # only if there is something before the space
 317                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 318                         # french spaces, Guillemet-right
 319                         '/(\\302\\253) /' => '\\1&nbsp;',
 320                 );
 321                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 322
 323                 # only once and last
 324                 $text = $this->doBlockLevels( $text, $linestart );
 325
 326                 $this->replaceLinkHolders( $text );
 327
 328                 # the position of the parserConvert() call should not be changed. it
 329                 # assumes that the links are all replaced and the only thing left
 330                 # is the <nowiki> mark.
 331                 # Side-effects: this calls $this->mOutput->setTitleText()
 332                 $text = $wgContLang->parserConvert( $text, $this );
 333
 334                 $text = $this->mStripState->unstripNoWiki( $text );
 335
 336                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 337
 338 //!JF Move to its own function
 339
 340                 $uniq_prefix = $this->mUniqPrefix;
 341                 $matches = array();
 342                 $elements = array_keys( $this->mTransparentTagHooks );
 343                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 344
 345                 foreach( $matches as $marker => $data ) {
 346                         list( $element, $content, $params, $tag ) = $data;
 347                         $tagName = strtolower( $element );
 348                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 349                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 350                                         array( $content, $params, $this ) );
 351                         } else {
 352                                 $output = $tag;
 353                         }
 354                         $this->mStripState->general->setPair( $marker, $output );
 355                 }
 356                 $text = $this->mStripState->unstripGeneral( $text );
 357
 358                 $text = Sanitizer::normalizeCharReferences( $text );
 359
 360                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 361                         $text = Parser::tidy($text);
 362                 } else {
 363                         # attempt to sanitize at least some nesting problems
 364                         # (bug #2702 and quite a few others)
 365                         $tidyregs = array(
 366                                 # ''Something [http://www.cool.com cool''] -->
 367                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 368                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 369                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 370                                 # fix up an anchor inside another anchor, only
 371                                 # at least for a single single nested link (bug 3695)
 372                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 373                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 374                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 375                                 # contains a div, so fix it up here; replace
 376                                 # div with escaped text
 377                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 378                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 379                                 # remove empty italic or bold tag pairs, some
 380                                 # introduced by rules above
 381                                 '/<([bi])><\/\\1>/' => '',
 382                         );
 383
 384                         $text = preg_replace(
 385                                 array_keys( $tidyregs ),
 386                                 array_values( $tidyregs ),
 387                                 $text );
 388                 }
 389
 390                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 391
 392                 # Information on include size limits, for the benefit of users who try to skirt them
 393                 if ( $this->mOptions->getEnableLimitReport() ) {
 394                         $max = $this->mOptions->getMaxIncludeSize();
 395                         $limitReport =
 396                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 397                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 398                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 399                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 400                         $text .= "\n<!-- \n$limitReport-->\n";
 401                 }
 402                 $this->mOutput->setText( $text );
 403                 $this->mRevisionId = $oldRevisionId;
 404                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 405                 wfProfileOut( $fname );
 406                 wfProfileOut( __METHOD__ );
 407
 408                 return $this->mOutput;
 409         }
 410
 411         /**
 412          * Recursive parser entry point that can be called from an extension tag
 413          * hook.
 414          */
 415         function recursiveTagParse( $text ) {
 416                 wfProfileIn( __METHOD__ );
 417                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 418                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 419                 $text = $this->internalParse( $text );
 420                 wfProfileOut( __METHOD__ );
 421                 return $text;
 422         }
 423
 424         /**
 425          * Expand templates and variables in the text, producing valid, static wikitext.
 426          * Also removes comments.
 427          */
 428         function preprocess( $text, $title, $options, $revid = null ) {
 429                 wfProfileIn( __METHOD__ );
 430                 $this->clearState();
 431                 $this->setOutputType( OT_PREPROCESS );
 432                 $this->mOptions = $options;
 433                 $this->mTitle = $title;
 434                 if( $revid !== null ) {
 435                         $this->mRevisionId = $revid;
 436                 }
 437                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 438                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 439                 $text = $this->replaceVariables( $text );
 440                 if ( $this->mOptions->getRemoveComments() ) {
 441                         $text = Sanitizer::removeHTMLcomments( $text );
 442                 }
 443                 $text = $this->mStripState->unstripBoth( $text );
 444                 wfProfileOut( __METHOD__ );
 445                 return $text;
 446         }
 447
 448         /**
 449          * Get a random string
 450          *
 451          * @private
 452          * @static
 453          */
 454         function getRandomString() {
 455                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 456         }
 457
 458         function &getTitle() { return $this->mTitle; }
 459         function getOptions() { return $this->mOptions; }
 460
 461         function getFunctionLang() {
 462                 global $wgLang, $wgContLang;
 463                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 464         }
 465
 466         /**
 467          * Replaces all occurrences of HTML-style comments and the given tags
 468          * in the text with a random marker and returns teh next text. The output
 469          * parameter $matches will be an associative array filled with data in
 470          * the form:
 471          *   'UNIQ-xxxxx' => array(
 472          *     'element',
 473          *     'tag content',
 474          *     array( 'param' => 'x' ),
 475          *     '<element param="x">tag content</element>' ) )
 476          *
 477          * @param $elements list of element names. Comments are always extracted.
 478          * @param $text Source text string.
 479          * @param $uniq_prefix
 480          *
 481          * @public
 482          * @static
 483          */
 484         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 485                 static $n = 1;
 486                 $stripped = '';
 487                 $matches = array();
 488
 489                 $taglist = implode( '|', $elements );
 490                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 491
 492                 while ( '' != $text ) {
 493                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 494                         $stripped .= $p[0];
 495                         if( count( $p ) < 5 ) {
 496                                 break;
 497                         }
 498                         if( count( $p ) > 5 ) {
 499                                 // comment
 500                                 $element    = $p[4];
 501                                 $attributes = '';
 502                                 $close      = '';
 503                                 $inside     = $p[5];
 504                         } else {
 505                                 // tag
 506                                 $element    = $p[1];
 507                                 $attributes = $p[2];
 508                                 $close      = $p[3];
 509                                 $inside     = $p[4];
 510                         }
 511
 512                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 513                         $stripped .= $marker;
 514
 515                         if ( $close === '/>' ) {
 516                                 // Empty element tag, <tag />
 517                                 $content = null;
 518                                 $text = $inside;
 519                                 $tail = null;
 520                         } else {
 521                                 if( $element == '!--' ) {
 522                                         $end = '/(-->)/';
 523                                 } else {
 524                                         $end = "/(<\\/$element\\s*>)/i";
 525                                 }
 526                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 527                                 $content = $q[0];
 528                                 if( count( $q ) < 3 ) {
 529                                         # No end tag -- let it run out to the end of the text.
 530                                         $tail = '';
 531                                         $text = '';
 532                                 } else {
 533                                         $tail = $q[1];
 534                                         $text = $q[2];
 535                                 }
 536                         }
 537
 538                         $matches[$marker] = array( $element,
 539                                 $content,
 540                                 Sanitizer::decodeTagAttributes( $attributes ),
 541                                 "<$element$attributes$close$content$tail" );
 542                 }
 543                 return $stripped;
 544         }
 545
 546         /**
 547          * Get a list of strippable XML-like elements
 548          */
 549         function getStripList() {
 550                 global $wgRawHtml;
 551                 $elements = $this->mStripList;
 552                 if( $wgRawHtml ) {
 553                         $elements[] = 'html';
 554                 }
 555                 if( $this->mOptions->getUseTeX() ) {
 556                         $elements[] = 'math';
 557                 }
 558                 return $elements;
 559         }
 560
 561         /**
 562          * @deprecated use replaceVariables
 563          */
 564         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 565                 return $text;
 566         }
 567
 568         /**
 569          * Restores pre, math, and other extensions removed by strip()
 570          *
 571          * always call unstripNoWiki() after this one
 572          * @private
 573          * @deprecated use $this->mStripState->unstrip()
 574          */
 575         function unstrip( $text, $state ) {
 576                 return $state->unstripGeneral( $text );
 577         }
 578
 579         /**
 580          * Always call this after unstrip() to preserve the order
 581          *
 582          * @private
 583          * @deprecated use $this->mStripState->unstrip()
 584          */
 585         function unstripNoWiki( $text, $state ) {
 586                 return $state->unstripNoWiki( $text );
 587         }
 588
 589         /**
 590          * @deprecated use $this->mStripState->unstripBoth()
 591          */
 592         function unstripForHTML( $text ) {
 593                 return $this->mStripState->unstripBoth( $text );
 594         }
 595
 596         /**
 597          * Add an item to the strip state
 598          * Returns the unique tag which must be inserted into the stripped text
 599          * The tag will be replaced with the original text in unstrip()
 600          *
 601          * @private
 602          */
 603         function insertStripItem( $text ) {
 604                 static $n = 0;
 605                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 606                 ++$n;
 607                 $this->mStripState->general->setPair( $rnd, $text );
 608                 return $rnd;
 609         }
 610
 611         /**
 612          * Interface with html tidy, used if $wgUseTidy = true.
 613          * If tidy isn't able to correct the markup, the original will be
 614          * returned in all its glory with a warning comment appended.
 615          *
 616          * Either the external tidy program or the in-process tidy extension
 617          * will be used depending on availability. Override the default
 618          * $wgTidyInternal setting to disable the internal if it's not working.
 619          *
 620          * @param string $text Hideous HTML input
 621          * @return string Corrected HTML output
 622          * @public
 623          * @static
 624          */
 625         function tidy( $text ) {
 626                 global $wgTidyInternal;
 627                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 628 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 629 '<head><title>test</title></head><body>'.$text.'</body></html>';
 630                 if( $wgTidyInternal ) {
 631                         $correctedtext = Parser::internalTidy( $wrappedtext );
 632                 } else {
 633                         $correctedtext = Parser::externalTidy( $wrappedtext );
 634                 }
 635                 if( is_null( $correctedtext ) ) {
 636                         wfDebug( "Tidy error detected!\n" );
 637                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 638                 }
 639                 return $correctedtext;
 640         }
 641
 642         /**
 643          * Spawn an external HTML tidy process and get corrected markup back from it.
 644          *
 645          * @private
 646          * @static
 647          */
 648         function externalTidy( $text ) {
 649                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 650                 $fname = 'Parser::externalTidy';
 651                 wfProfileIn( $fname );
 652
 653                 $cleansource = '';
 654                 $opts = ' -utf8';
 655
 656                 $descriptorspec = array(
 657                         0 => array('pipe', 'r'),
 658                         1 => array('pipe', 'w'),
 659                         2 => array('file', wfGetNull(), 'a')
 660                 );
 661                 $pipes = array();
 662                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 663                 if (is_resource($process)) {
 664                         // Theoretically, this style of communication could cause a deadlock
 665                         // here. If the stdout buffer fills up, then writes to stdin could
 666                         // block. This doesn't appear to happen with tidy, because tidy only
 667                         // writes to stdout after it's finished reading from stdin. Search
 668                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 669                         fwrite($pipes[0], $text);
 670                         fclose($pipes[0]);
 671                         while (!feof($pipes[1])) {
 672                                 $cleansource .= fgets($pipes[1], 1024);
 673                         }
 674                         fclose($pipes[1]);
 675                         proc_close($process);
 676                 }
 677
 678                 wfProfileOut( $fname );
 679
 680                 if( $cleansource == '' && $text != '') {
 681                         // Some kind of error happened, so we couldn't get the corrected text.
 682                         // Just give up; we'll use the source text and append a warning.
 683                         return null;
 684                 } else {
 685                         return $cleansource;
 686                 }
 687         }
 688
 689         /**
 690          * Use the HTML tidy PECL extension to use the tidy library in-process,
 691          * saving the overhead of spawning a new process.
 692          *
 693          * 'pear install tidy' should be able to compile the extension module.
 694          *
 695          * @private
 696          * @static
 697          */
 698         function internalTidy( $text ) {
 699                 global $wgTidyConf, $IP, $wgDebugTidy;
 700                 $fname = 'Parser::internalTidy';
 701                 wfProfileIn( $fname );
 702
 703                 $tidy = new tidy;
 704                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 705                 $tidy->cleanRepair();
 706                 if( $tidy->getStatus() == 2 ) {
 707                         // 2 is magic number for fatal error
 708                         // http://www.php.net/manual/en/function.tidy-get-status.php
 709                         $cleansource = null;
 710                 } else {
 711                         $cleansource = tidy_get_output( $tidy );
 712                 }
 713                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 714                         $cleansource .= "<!--\nTidy reports:\n" .
 715                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 716                                 "\n-->";
 717                 }
 718
 719                 wfProfileOut( $fname );
 720                 return $cleansource;
 721         }
 722
 723         /**
 724          * parse the wiki syntax used to render tables
 725          *
 726          * @private
 727          */
 728         function doTableStuff ( $text ) {
 729                 $fname = 'Parser::doTableStuff';
 730                 wfProfileIn( $fname );
 731
 732                 $lines = explode ( "\n" , $text );
 733                 $td_history = array (); // Is currently a td tag open?
 734                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 735                 $tr_history = array (); // Is currently a tr tag open?
 736                 $tr_attributes = array (); // history of tr attributes
 737                 $has_opened_tr = array(); // Did this table open a <tr> element?
 738                 $indent_level = 0; // indent level of the table
 739                 foreach ( $lines as $key => $line )
 740                 {
 741                         $line = trim ( $line );
 742
 743                         if( $line == '' ) { // empty line, go to next line
 744                                 continue;
 745                         }
 746                         $first_character = $line{0};
 747                         $matches = array();
 748
 749                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 750                                 // First check if we are starting a new table
 751                                 $indent_level = strlen( $matches[1] );
 752
 753                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 754                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 755
 756                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 757                                 array_push ( $td_history , false );
 758                                 array_push ( $last_tag_history , '' );
 759                                 array_push ( $tr_history , false );
 760                                 array_push ( $tr_attributes , '' );
 761                                 array_push ( $has_opened_tr , false );
 762                         } else if ( count ( $td_history ) == 0 ) {
 763                                 // Don't do any of the following
 764                                 continue;
 765                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 766                                 // We are ending a table
 767                                 $line = '</table>' . substr ( $line , 2 );
 768                                 $last_tag = array_pop ( $last_tag_history );
 769
 770                                 if ( !array_pop ( $has_opened_tr ) ) {
 771                                         $line = "<tr><td></td></tr>{$line}";
 772                                 }
 773
 774                                 if ( array_pop ( $tr_history ) ) {
 775                                         $line = "</tr>{$line}";
 776                                 }
 777
 778                                 if ( array_pop ( $td_history ) ) {
 779                                         $line = "</{$last_tag}>{$line}";
 780                                 }
 781                                 array_pop ( $tr_attributes );
 782                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 783                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 784                                 // Now we have a table row
 785                                 $line = preg_replace( '#^\|-+#', '', $line );
 786
 787                                 // Whats after the tag is now only attributes
 788                                 $attributes = $this->mStripState->unstripBoth( $line );
 789                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 790                                 array_pop ( $tr_attributes );
 791                                 array_push ( $tr_attributes , $attributes );
 792
 793                                 $line = '';
 794                                 $last_tag = array_pop ( $last_tag_history );
 795                                 array_pop ( $has_opened_tr );
 796                                 array_push ( $has_opened_tr , true );
 797
 798                                 if ( array_pop ( $tr_history ) ) {
 799                                         $line = '</tr>';
 800                                 }
 801
 802                                 if ( array_pop ( $td_history ) ) {
 803                                         $line = "</{$last_tag}>{$line}";
 804                                 }
 805
 806                                 $lines[$key] = $line;
 807                                 array_push ( $tr_history , false );
 808                                 array_push ( $td_history , false );
 809                                 array_push ( $last_tag_history , '' );
 810                         }
 811                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 812                                 // This might be cell elements, td, th or captions
 813                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 814                                         $first_character = '+';
 815                                         $line = substr ( $line , 1 );
 816                                 }
 817
 818                                 $line = substr ( $line , 1 );
 819
 820                                 if ( $first_character == '!' ) {
 821                                         $line = str_replace ( '!!' , '||' , $line );
 822                                 }
 823
 824                                 // Split up multiple cells on the same line.
 825                                 // FIXME : This can result in improper nesting of tags processed
 826                                 // by earlier parser steps, but should avoid splitting up eg
 827                                 // attribute values containing literal "||".
 828                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 829
 830                                 $lines[$key] = '';
 831
 832                                 // Loop through each table cell
 833                                 foreach ( $cells as $cell )
 834                                 {
 835                                         $previous = '';
 836                                         if ( $first_character != '+' )
 837                                         {
 838                                                 $tr_after = array_pop ( $tr_attributes );
 839                                                 if ( !array_pop ( $tr_history ) ) {
 840                                                         $previous = "<tr{$tr_after}>\n";
 841                                                 }
 842                                                 array_push ( $tr_history , true );
 843                                                 array_push ( $tr_attributes , '' );
 844                                                 array_pop ( $has_opened_tr );
 845                                                 array_push ( $has_opened_tr , true );
 846                                         }
 847
 848                                         $last_tag = array_pop ( $last_tag_history );
 849
 850                                         if ( array_pop ( $td_history ) ) {
 851                                                 $previous = "</{$last_tag}>{$previous}";
 852                                         }
 853
 854                                         if ( $first_character == '|' ) {
 855                                                 $last_tag = 'td';
 856                                         } else if ( $first_character == '!' ) {
 857                                                 $last_tag = 'th';
 858                                         } else if ( $first_character == '+' ) {
 859                                                 $last_tag = 'caption';
 860                                         } else {
 861                                                 $last_tag = '';
 862                                         }
 863
 864                                         array_push ( $last_tag_history , $last_tag );
 865
 866                                         // A cell could contain both parameters and data
 867                                         $cell_data = explode ( '|' , $cell , 2 );
 868
 869                                         // Bug 553: Note that a '|' inside an invalid link should not
 870                                         // be mistaken as delimiting cell parameters
 871                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 872                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 873                                         } else if ( count ( $cell_data ) == 1 )
 874                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 875                                         else {
 876                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 877                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 878                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 879                                         }
 880
 881                                         $lines[$key] .= $cell;
 882                                         array_push ( $td_history , true );
 883                                 }
 884                         }
 885                 }
 886
 887                 // Closing open td, tr && table
 888                 while ( count ( $td_history ) > 0 )
 889                 {
 890                         if ( array_pop ( $td_history ) ) {
 891                                 $lines[] = '</td>' ;
 892                         }
 893                         if ( array_pop ( $tr_history ) ) {
 894                                 $lines[] = '</tr>' ;
 895                         }
 896                         if ( !array_pop ( $has_opened_tr ) ) {
 897                                 $lines[] = "<tr><td></td></tr>" ;
 898                         }
 899
 900                         $lines[] = '</table>' ;
 901                 }
 902
 903                 $output = implode ( "\n" , $lines ) ;
 904
 905                 // special case: don't return empty table
 906                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 907                         $output = '';
 908                 }
 909
 910                 wfProfileOut( $fname );
 911
 912                 return $output;
 913         }
 914
 915         /**
 916          * Helper function for parse() that transforms wiki markup into
 917          * HTML. Only called for $mOutputType == OT_HTML.
 918          *
 919          * @private
 920          */
 921         function internalParse( $text ) {
 922                 $isMain = true;
 923                 $fname = 'Parser::internalParse';
 924                 wfProfileIn( $fname );
 925
 926                 # Hook to suspend the parser in this state
 927                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 928                         wfProfileOut( $fname );
 929                         return $text ;
 930                 }
 931
 932                 $text = $this->replaceVariables( $text );
 933                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 934                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 935
 936                 // Tables need to come after variable replacement for things to work
 937                 // properly; putting them before other transformations should keep
 938                 // exciting things like link expansions from showing up in surprising
 939                 // places.
 940                 $text = $this->doTableStuff( $text );
 941
 942                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 943
 944                 $text = $this->stripToc( $text );
 945                 $this->stripNoGallery( $text );
 946                 $text = $this->doHeadings( $text );
 947                 if($this->mOptions->getUseDynamicDates()) {
 948                         $df =& DateFormatter::getInstance();
 949                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 950                 }
 951                 $text = $this->doAllQuotes( $text );
 952                 $text = $this->replaceInternalLinks( $text );
 953                 $text = $this->replaceExternalLinks( $text );
 954
 955                 # replaceInternalLinks may sometimes leave behind
 956                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 957                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 958
 959                 $text = $this->doMagicLinks( $text );
 960                 $text = $this->formatHeadings( $text, $isMain );
 961
 962                 wfProfileOut( $fname );
 963                 return $text;
 964         }
 965
 966         /**
 967          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 968          * magic external links.
 969          *
 970          * @private
 971          */
 972         function doMagicLinks( $text ) {
 973                 wfProfileIn( __METHOD__ );
 974                 $text = preg_replace_callback(
 975                         '!(?:                           # Start cases
 976                             <a.*?</a> |                 # Skip link text
 977                             <.*?> |                     # Skip stuff inside HTML elements
 978                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 979                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 980                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 981                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 982                                      [0-9Xx]                 # check digit
 983                                    \b)
 984                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 985                 wfProfileOut( __METHOD__ );
 986                 return $text;
 987         }
 988
 989         function magicLinkCallback( $m ) {
 990                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 991                         # Skip HTML element
 992                         return $m[0];
 993                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 994                         $isbn = $m[2];
 995                         $num = strtr( $isbn, array(
 996                                 '-' => '',
 997                                 ' ' => '',
 998                                 'x' => 'X',
 999                         ));
1000                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1001                         $text = '<a href="' .
1002                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1003                                 "\" class=\"internal\">ISBN $isbn</a>";
1004                 } else {
1005                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1006                                 $keyword = 'RFC';
1007                                 $urlmsg = 'rfcurl';
1008                                 $id = $m[1];
1009                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1010                                 $keyword = 'PMID';
1011                                 $urlmsg = 'pubmedurl';
1012                                 $id = $m[1];
1013                         } else {
1014                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1015                                         substr($m[0], 0, 20 ) . '"' );
1016                         }
1017
1018                         $url = wfMsg( $urlmsg, $id);
1019                         $sk = $this->mOptions->getSkin();
1020                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1021                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1022                 }
1023                 return $text;
1024         }
1025
1026         /**
1027          * Parse headers and return html
1028          *
1029          * @private
1030          */
1031         function doHeadings( $text ) {
1032                 $fname = 'Parser::doHeadings';
1033                 wfProfileIn( $fname );
1034                 for ( $i = 6; $i >= 1; --$i ) {
1035                         $h = str_repeat( '=', $i );
1036                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1037                           "<h$i>\\1</h$i>", $text );
1038                 }
1039                 wfProfileOut( $fname );
1040                 return $text;
1041         }
1042
1043         /**
1044          * Replace single quotes with HTML markup
1045          * @private
1046          * @return string the altered text
1047          */
1048         function doAllQuotes( $text ) {
1049                 $fname = 'Parser::doAllQuotes';
1050                 wfProfileIn( $fname );
1051                 $outtext = '';
1052                 $lines = explode( "\n", $text );
1053                 foreach ( $lines as $line ) {
1054                         $outtext .= $this->doQuotes ( $line ) . "\n";
1055                 }
1056                 $outtext = substr($outtext, 0,-1);
1057                 wfProfileOut( $fname );
1058                 return $outtext;
1059         }
1060
1061         /**
1062          * Helper function for doAllQuotes()
1063          */
1064         public function doQuotes( $text ) {
1065                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1066                 if ( count( $arr ) == 1 )
1067                         return $text;
1068                 else
1069                 {
1070                         # First, do some preliminary work. This may shift some apostrophes from
1071                         # being mark-up to being text. It also counts the number of occurrences
1072                         # of bold and italics mark-ups.
1073                         $i = 0;
1074                         $numbold = 0;
1075                         $numitalics = 0;
1076                         foreach ( $arr as $r )
1077                         {
1078                                 if ( ( $i % 2 ) == 1 )
1079                                 {
1080                                         # If there are ever four apostrophes, assume the first is supposed to
1081                                         # be text, and the remaining three constitute mark-up for bold text.
1082                                         if ( strlen( $arr[$i] ) == 4 )
1083                                         {
1084                                                 $arr[$i-1] .= "'";
1085                                                 $arr[$i] = "'''";
1086                                         }
1087                                         # If there are more than 5 apostrophes in a row, assume they're all
1088                                         # text except for the last 5.
1089                                         else if ( strlen( $arr[$i] ) > 5 )
1090                                         {
1091                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1092                                                 $arr[$i] = "'''''";
1093                                         }
1094                                         # Count the number of occurrences of bold and italics mark-ups.
1095                                         # We are not counting sequences of five apostrophes.
1096                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1097                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1098                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1099                                 }
1100                                 $i++;
1101                         }
1102
1103                         # If there is an odd number of both bold and italics, it is likely
1104                         # that one of the bold ones was meant to be an apostrophe followed
1105                         # by italics. Which one we cannot know for certain, but it is more
1106                         # likely to be one that has a single-letter word before it.
1107                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1108                         {
1109                                 $i = 0;
1110                                 $firstsingleletterword = -1;
1111                                 $firstmultiletterword = -1;
1112                                 $firstspace = -1;
1113                                 foreach ( $arr as $r )
1114                                 {
1115                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1116                                         {
1117                                                 $x1 = substr ($arr[$i-1], -1);
1118                                                 $x2 = substr ($arr[$i-1], -2, 1);
1119                                                 if ($x1 == ' ') {
1120                                                         if ($firstspace == -1) $firstspace = $i;
1121                                                 } else if ($x2 == ' ') {
1122                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1123                                                 } else {
1124                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1125                                                 }
1126                                         }
1127                                         $i++;
1128                                 }
1129
1130                                 # If there is a single-letter word, use it!
1131                                 if ($firstsingleletterword > -1)
1132                                 {
1133                                         $arr [ $firstsingleletterword ] = "''";
1134                                         $arr [ $firstsingleletterword-1 ] .= "'";
1135                                 }
1136                                 # If not, but there's a multi-letter word, use that one.
1137                                 else if ($firstmultiletterword > -1)
1138                                 {
1139                                         $arr [ $firstmultiletterword ] = "''";
1140                                         $arr [ $firstmultiletterword-1 ] .= "'";
1141                                 }
1142                                 # ... otherwise use the first one that has neither.
1143                                 # (notice that it is possible for all three to be -1 if, for example,
1144                                 # there is only one pentuple-apostrophe in the line)
1145                                 else if ($firstspace > -1)
1146                                 {
1147                                         $arr [ $firstspace ] = "''";
1148                                         $arr [ $firstspace-1 ] .= "'";
1149                                 }
1150                         }
1151
1152                         # Now let's actually convert our apostrophic mush to HTML!
1153                         $output = '';
1154                         $buffer = '';
1155                         $state = '';
1156                         $i = 0;
1157                         foreach ($arr as $r)
1158                         {
1159                                 if (($i % 2) == 0)
1160                                 {
1161                                         if ($state == 'both')
1162                                                 $buffer .= $r;
1163                                         else
1164                                                 $output .= $r;
1165                                 }
1166                                 else
1167                                 {
1168                                         if (strlen ($r) == 2)
1169                                         {
1170                                                 if ($state == 'i')
1171                                                 { $output .= '</i>'; $state = ''; }
1172                                                 else if ($state == 'bi')
1173                                                 { $output .= '</i>'; $state = 'b'; }
1174                                                 else if ($state == 'ib')
1175                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1176                                                 else if ($state == 'both')
1177                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1178                                                 else # $state can be 'b' or ''
1179                                                 { $output .= '<i>'; $state .= 'i'; }
1180                                         }
1181                                         else if (strlen ($r) == 3)
1182                                         {
1183                                                 if ($state == 'b')
1184                                                 { $output .= '</b>'; $state = ''; }
1185                                                 else if ($state == 'bi')
1186                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1187                                                 else if ($state == 'ib')
1188                                                 { $output .= '</b>'; $state = 'i'; }
1189                                                 else if ($state == 'both')
1190                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1191                                                 else # $state can be 'i' or ''
1192                                                 { $output .= '<b>'; $state .= 'b'; }
1193                                         }
1194                                         else if (strlen ($r) == 5)
1195                                         {
1196                                                 if ($state == 'b')
1197                                                 { $output .= '</b><i>'; $state = 'i'; }
1198                                                 else if ($state == 'i')
1199                                                 { $output .= '</i><b>'; $state = 'b'; }
1200                                                 else if ($state == 'bi')
1201                                                 { $output .= '</i></b>'; $state = ''; }
1202                                                 else if ($state == 'ib')
1203                                                 { $output .= '</b></i>'; $state = ''; }
1204                                                 else if ($state == 'both')
1205                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1206                                                 else # ($state == '')
1207                                                 { $buffer = ''; $state = 'both'; }
1208                                         }
1209                                 }
1210                                 $i++;
1211                         }
1212                         # Now close all remaining tags.  Notice that the order is important.
1213                         if ($state == 'b' || $state == 'ib')
1214                                 $output .= '</b>';
1215                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1216                                 $output .= '</i>';
1217                         if ($state == 'bi')
1218                                 $output .= '</b>';
1219                         # There might be lonely ''''', so make sure we have a buffer
1220                         if ($state == 'both' && $buffer)
1221                                 $output .= '<b><i>'.$buffer.'</i></b>';
1222                         return $output;
1223                 }
1224         }
1225
1226         /**
1227          * Replace external links
1228          *
1229          * Note: this is all very hackish and the order of execution matters a lot.
1230          * Make sure to run maintenance/parserTests.php if you change this code.
1231          *
1232          * @private
1233          */
1234         function replaceExternalLinks( $text ) {
1235                 global $wgContLang;
1236                 $fname = 'Parser::replaceExternalLinks';
1237                 wfProfileIn( $fname );
1238
1239                 $sk = $this->mOptions->getSkin();
1240
1241                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1242
1243                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1244
1245                 $i = 0;
1246                 while ( $i<count( $bits ) ) {
1247                         $url = $bits[$i++];
1248                         $protocol = $bits[$i++];
1249                         $text = $bits[$i++];
1250                         $trail = $bits[$i++];
1251
1252                         # The characters '<' and '>' (which were escaped by
1253                         # removeHTMLtags()) should not be included in
1254                         # URLs, per RFC 2396.
1255                         $m2 = array();
1256                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1257                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1258                                 $url = substr($url, 0, $m2[0][1]);
1259                         }
1260
1261                         # If the link text is an image URL, replace it with an <img> tag
1262                         # This happened by accident in the original parser, but some people used it extensively
1263                         $img = $this->maybeMakeExternalImage( $text );
1264                         if ( $img !== false ) {
1265                                 $text = $img;
1266                         }
1267
1268                         $dtrail = '';
1269
1270                         # Set linktype for CSS - if URL==text, link is essentially free
1271                         $linktype = ($text == $url) ? 'free' : 'text';
1272
1273                         # No link text, e.g. [http://domain.tld/some.link]
1274                         if ( $text == '' ) {
1275                                 # Autonumber if allowed. See bug #5918
1276                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1277                                         $text = '[' . ++$this->mAutonumber . ']';
1278                                         $linktype = 'autonumber';
1279                                 } else {
1280                                         # Otherwise just use the URL
1281                                         $text = htmlspecialchars( $url );
1282                                         $linktype = 'free';
1283                                 }
1284                         } else {
1285                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1286                                 # Check for trail
1287                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1288                         }
1289
1290                         $text = $wgContLang->markNoConversion($text);
1291
1292                         $url = Sanitizer::cleanUrl( $url );
1293
1294                         # Process the trail (i.e. everything after this link up until start of the next link),
1295                         # replacing any non-bracketed links
1296                         $trail = $this->replaceFreeExternalLinks( $trail );
1297
1298                         # Use the encoded URL
1299                         # This means that users can paste URLs directly into the text
1300                         # Funny characters like &ouml; aren't valid in URLs anyway
1301                         # This was changed in August 2004
1302                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1303
1304                         # Register link in the output object.
1305                         # Replace unnecessary URL escape codes with the referenced character
1306                         # This prevents spammers from hiding links from the filters
1307                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1308                         $this->mOutput->addExternalLink( $pasteurized );
1309                 }
1310
1311                 wfProfileOut( $fname );
1312                 return $s;
1313         }
1314
1315         /**
1316          * Replace anything that looks like a URL with a link
1317          * @private
1318          */
1319         function replaceFreeExternalLinks( $text ) {
1320                 global $wgContLang;
1321                 $fname = 'Parser::replaceFreeExternalLinks';
1322                 wfProfileIn( $fname );
1323
1324                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1325                 $s = array_shift( $bits );
1326                 $i = 0;
1327
1328                 $sk = $this->mOptions->getSkin();
1329
1330                 while ( $i < count( $bits ) ){
1331                         $protocol = $bits[$i++];
1332                         $remainder = $bits[$i++];
1333
1334                         $m = array();
1335                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1336                                 # Found some characters after the protocol that look promising
1337                                 $url = $protocol . $m[1];
1338                                 $trail = $m[2];
1339
1340                                 # special case: handle urls as url args:
1341                                 # http://www.example.com/foo?=http://www.example.com/bar
1342                                 if(strlen($trail) == 0 &&
1343                                         isset($bits[$i]) &&
1344                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1345                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1346                                 {
1347                                         # add protocol, arg
1348                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1349                                         $i += 2;
1350                                         $trail = $m[2];
1351                                 }
1352
1353                                 # The characters '<' and '>' (which were escaped by
1354                                 # removeHTMLtags()) should not be included in
1355                                 # URLs, per RFC 2396.
1356                                 $m2 = array();
1357                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1358                                         $trail = substr($url, $m2[0][1]) . $trail;
1359                                         $url = substr($url, 0, $m2[0][1]);
1360                                 }
1361
1362                                 # Move trailing punctuation to $trail
1363                                 $sep = ',;\.:!?';
1364                                 # If there is no left bracket, then consider right brackets fair game too
1365                                 if ( strpos( $url, '(' ) === false ) {
1366                                         $sep .= ')';
1367                                 }
1368
1369                                 $numSepChars = strspn( strrev( $url ), $sep );
1370                                 if ( $numSepChars ) {
1371                                         $trail = substr( $url, -$numSepChars ) . $trail;
1372                                         $url = substr( $url, 0, -$numSepChars );
1373                                 }
1374
1375                                 $url = Sanitizer::cleanUrl( $url );
1376
1377                                 # Is this an external image?
1378                                 $text = $this->maybeMakeExternalImage( $url );
1379                                 if ( $text === false ) {
1380                                         # Not an image, make a link
1381                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1382                                         # Register it in the output object...
1383                                         # Replace unnecessary URL escape codes with their equivalent characters
1384                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1385                                         $this->mOutput->addExternalLink( $pasteurized );
1386                                 }
1387                                 $s .= $text . $trail;
1388                         } else {
1389                                 $s .= $protocol . $remainder;
1390                         }
1391                 }
1392                 wfProfileOut( $fname );
1393                 return $s;
1394         }
1395
1396         /**
1397          * Replace unusual URL escape codes with their equivalent characters
1398          * @param string
1399          * @return string
1400          * @static
1401          * @todo  This can merge genuinely required bits in the path or query string,
1402          *        breaking legit URLs. A proper fix would treat the various parts of
1403          *        the URL differently; as a workaround, just use the output for
1404          *        statistical records, not for actual linking/output.
1405          */
1406         static function replaceUnusualEscapes( $url ) {
1407                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1408                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1409         }
1410
1411         /**
1412          * Callback function used in replaceUnusualEscapes().
1413          * Replaces unusual URL escape codes with their equivalent character
1414          * @static
1415          * @private
1416          */
1417         private static function replaceUnusualEscapesCallback( $matches ) {
1418                 $char = urldecode( $matches[0] );
1419                 $ord = ord( $char );
1420                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1421                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1422                         // No, shouldn't be escaped
1423                         return $char;
1424                 } else {
1425                         // Yes, leave it escaped
1426                         return $matches[0];
1427                 }
1428         }
1429
1430         /**
1431          * make an image if it's allowed, either through the global
1432          * option or through the exception
1433          * @private
1434          */
1435         function maybeMakeExternalImage( $url ) {
1436                 $sk = $this->mOptions->getSkin();
1437                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1438                 $imagesexception = !empty($imagesfrom);
1439                 $text = false;
1440                 if ( $this->mOptions->getAllowExternalImages()
1441                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1442                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1443                                 # Image found
1444                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1445                         }
1446                 }
1447                 return $text;
1448         }
1449
1450         /**
1451          * Process [[ ]] wikilinks
1452          *
1453          * @private
1454          */
1455         function replaceInternalLinks( $s ) {
1456                 global $wgContLang;
1457                 static $fname = 'Parser::replaceInternalLinks' ;
1458
1459                 wfProfileIn( $fname );
1460
1461                 wfProfileIn( $fname.'-setup' );
1462                 static $tc = FALSE;
1463                 # the % is needed to support urlencoded titles as well
1464                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1465
1466                 $sk = $this->mOptions->getSkin();
1467
1468                 #split the entire text string on occurences of [[
1469                 $a = explode( '[[', ' ' . $s );
1470                 #get the first element (all text up to first [[), and remove the space we added
1471                 $s = array_shift( $a );
1472                 $s = substr( $s, 1 );
1473
1474                 # Match a link having the form [[namespace:link|alternate]]trail
1475                 static $e1 = FALSE;
1476                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1477                 # Match cases where there is no "]]", which might still be images
1478                 static $e1_img = FALSE;
1479                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1480
1481                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1482                 $e2 = null;
1483                 if ( $useLinkPrefixExtension ) {
1484                         # Match the end of a line for a word that's not followed by whitespace,
1485                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1486                         $e2 = wfMsgForContent( 'linkprefix' );
1487                 }
1488
1489                 if( is_null( $this->mTitle ) ) {
1490                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1491                 }
1492                 $nottalk = !$this->mTitle->isTalkPage();
1493
1494                 if ( $useLinkPrefixExtension ) {
1495                         $m = array();
1496                         if ( preg_match( $e2, $s, $m ) ) {
1497                                 $first_prefix = $m[2];
1498                         } else {
1499                                 $first_prefix = false;
1500                         }
1501                 } else {
1502                         $prefix = '';
1503                 }
1504
1505                 if($wgContLang->hasVariants()) {
1506                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1507                 } else {
1508                         $selflink = array($this->mTitle->getPrefixedText());
1509                 }
1510                 $useSubpages = $this->areSubpagesAllowed();
1511                 wfProfileOut( $fname.'-setup' );
1512
1513                 # Loop for each link
1514                 for ($k = 0; isset( $a[$k] ); $k++) {
1515                         $line = $a[$k];
1516                         if ( $useLinkPrefixExtension ) {
1517                                 wfProfileIn( $fname.'-prefixhandling' );
1518                                 if ( preg_match( $e2, $s, $m ) ) {
1519                                         $prefix = $m[2];
1520                                         $s = $m[1];
1521                                 } else {
1522                                         $prefix='';
1523                                 }
1524                                 # first link
1525                                 if($first_prefix) {
1526                                         $prefix = $first_prefix;
1527                                         $first_prefix = false;
1528                                 }
1529                                 wfProfileOut( $fname.'-prefixhandling' );
1530                         }
1531
1532                         $might_be_img = false;
1533
1534                         wfProfileIn( "$fname-e1" );
1535                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1536                                 $text = $m[2];
1537                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1538                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1539                                 # the real problem is with the $e1 regex
1540                                 # See bug 1300.
1541                                 #
1542                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1543                                 # and no image is in sight. See bug 2095.
1544                                 #
1545                                 if( $text !== '' &&
1546                                         substr( $m[3], 0, 1 ) === ']' &&
1547                                         strpos($text, '[') !== false
1548                                 )
1549                                 {
1550                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1551                                         $m[3] = substr( $m[3], 1 );
1552                                 }
1553                                 # fix up urlencoded title texts
1554                                 if( strpos( $m[1], '%' ) !== false ) {
1555                                         # Should anchors '#' also be rejected?
1556                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1557                                 }
1558                                 $trail = $m[3];
1559                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1560                                 $might_be_img = true;
1561                                 $text = $m[2];
1562                                 if ( strpos( $m[1], '%' ) !== false ) {
1563                                         $m[1] = urldecode($m[1]);
1564                                 }
1565                                 $trail = "";
1566                         } else { # Invalid form; output directly
1567                                 $s .= $prefix . '[[' . $line ;
1568                                 wfProfileOut( "$fname-e1" );
1569                                 continue;
1570                         }
1571                         wfProfileOut( "$fname-e1" );
1572                         wfProfileIn( "$fname-misc" );
1573
1574                         # Don't allow internal links to pages containing
1575                         # PROTO: where PROTO is a valid URL protocol; these
1576                         # should be external links.
1577                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1578                                 $s .= $prefix . '[[' . $line ;
1579                                 continue;
1580                         }
1581
1582                         # Make subpage if necessary
1583                         if( $useSubpages ) {
1584                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1585                         } else {
1586                                 $link = $m[1];
1587                         }
1588
1589                         $noforce = (substr($m[1], 0, 1) != ':');
1590                         if (!$noforce) {
1591                                 # Strip off leading ':'
1592                                 $link = substr($link, 1);
1593                         }
1594
1595                         wfProfileOut( "$fname-misc" );
1596                         wfProfileIn( "$fname-title" );
1597                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1598                         if( !$nt ) {
1599                                 $s .= $prefix . '[[' . $line;
1600                                 wfProfileOut( "$fname-title" );
1601                                 continue;
1602                         }
1603
1604                         $ns = $nt->getNamespace();
1605                         $iw = $nt->getInterWiki();
1606                         wfProfileOut( "$fname-title" );
1607
1608                         if ($might_be_img) { # if this is actually an invalid link
1609                                 wfProfileIn( "$fname-might_be_img" );
1610                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1611                                         $found = false;
1612                                         while (isset ($a[$k+1]) ) {
1613                                                 #look at the next 'line' to see if we can close it there
1614                                                 $spliced = array_splice( $a, $k + 1, 1 );
1615                                                 $next_line = array_shift( $spliced );
1616                                                 $m = explode( ']]', $next_line, 3 );
1617                                                 if ( count( $m ) == 3 ) {
1618                                                         # the first ]] closes the inner link, the second the image
1619                                                         $found = true;
1620                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1621                                                         $trail = $m[2];
1622                                                         break;
1623                                                 } elseif ( count( $m ) == 2 ) {
1624                                                         #if there's exactly one ]] that's fine, we'll keep looking
1625                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1626                                                 } else {
1627                                                         #if $next_line is invalid too, we need look no further
1628                                                         $text .= '[[' . $next_line;
1629                                                         break;
1630                                                 }
1631                                         }
1632                                         if ( !$found ) {
1633                                                 # we couldn't find the end of this imageLink, so output it raw
1634                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1635                                                 $text = $this->replaceInternalLinks($text);
1636                                                 $s .= "{$prefix}[[$link|$text";
1637                                                 # note: no $trail, because without an end, there *is* no trail
1638                                                 wfProfileOut( "$fname-might_be_img" );
1639                                                 continue;
1640                                         }
1641                                 } else { #it's not an image, so output it raw
1642                                         $s .= "{$prefix}[[$link|$text";
1643                                         # note: no $trail, because without an end, there *is* no trail
1644                                         wfProfileOut( "$fname-might_be_img" );
1645                                         continue;
1646                                 }
1647                                 wfProfileOut( "$fname-might_be_img" );
1648                         }
1649
1650                         $wasblank = ( '' == $text );
1651                         if( $wasblank ) $text = $link;
1652
1653                         # Link not escaped by : , create the various objects
1654                         if( $noforce ) {
1655
1656                                 # Interwikis
1657                                 wfProfileIn( "$fname-interwiki" );
1658                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1659                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1660                                         $s = rtrim($s . $prefix);
1661                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1662                                         wfProfileOut( "$fname-interwiki" );
1663                                         continue;
1664                                 }
1665                                 wfProfileOut( "$fname-interwiki" );
1666
1667                                 if ( $ns == NS_IMAGE ) {
1668                                         wfProfileIn( "$fname-image" );
1669                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1670                                                 # recursively parse links inside the image caption
1671                                                 # actually, this will parse them in any other parameters, too,
1672                                                 # but it might be hard to fix that, and it doesn't matter ATM
1673                                                 $text = $this->replaceExternalLinks($text);
1674                                                 $text = $this->replaceInternalLinks($text);
1675
1676                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1677                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1678                                                 $this->mOutput->addImage( $nt->getDBkey() );
1679
1680                                                 wfProfileOut( "$fname-image" );
1681                                                 continue;
1682                                         } else {
1683                                                 # We still need to record the image's presence on the page
1684                                                 $this->mOutput->addImage( $nt->getDBkey() );
1685                                         }
1686                                         wfProfileOut( "$fname-image" );
1687
1688                                 }
1689
1690                                 if ( $ns == NS_CATEGORY ) {
1691                                         wfProfileIn( "$fname-category" );
1692                                         $s = rtrim($s . "\n"); # bug 87
1693
1694                                         if ( $wasblank ) {
1695                                                 $sortkey = $this->getDefaultSort();
1696                                         } else {
1697                                                 $sortkey = $text;
1698                                         }
1699                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1700                                         $sortkey = str_replace( "\n", '', $sortkey );
1701                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1702                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1703
1704                                         /**
1705                                          * Strip the whitespace Category links produce, see bug 87
1706                                          * @todo We might want to use trim($tmp, "\n") here.
1707                                          */
1708                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1709
1710                                         wfProfileOut( "$fname-category" );
1711                                         continue;
1712                                 }
1713                         }
1714
1715                         # Self-link checking
1716                         if( $nt->getFragment() === '' ) {
1717                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1718                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1719                                         continue;
1720                                 }
1721                         }
1722
1723                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1724                         if( $ns == NS_MEDIA ) {
1725                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1726                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1727                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1728                                 $this->mOutput->addImage( $nt->getDBkey() );
1729                                 continue;
1730                         } elseif( $ns == NS_SPECIAL ) {
1731                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1732                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1733                                 } else {
1734                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1735                                 }
1736                                 continue;
1737                         } elseif( $ns == NS_IMAGE ) {
1738                                 $img = wfFindFile( $nt );
1739                                 if( $img ) {
1740                                         // Force a blue link if the file exists; may be a remote
1741                                         // upload on the shared repository, and we want to see its
1742                                         // auto-generated page.
1743                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1744                                         $this->mOutput->addLink( $nt );
1745                                         continue;
1746                                 }
1747                         }
1748                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1749                 }
1750                 wfProfileOut( $fname );
1751                 return $s;
1752         }
1753
1754         /**
1755          * Make a link placeholder. The text returned can be later resolved to a real link with
1756          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1757          * parsing of interwiki links, and secondly to allow all existence checks and
1758          * article length checks (for stub links) to be bundled into a single query.
1759          *
1760          */
1761         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1762                 wfProfileIn( __METHOD__ );
1763                 if ( ! is_object($nt) ) {
1764                         # Fail gracefully
1765                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1766                 } else {
1767                         # Separate the link trail from the rest of the link
1768                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1769
1770                         if ( $nt->isExternal() ) {
1771                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1772                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1773                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1774                         } else {
1775                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1776                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1777                                 $this->mLinkHolders['queries'][] = $query;
1778                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1779                                 $this->mLinkHolders['titles'][] = $nt;
1780
1781                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1782                         }
1783                 }
1784                 wfProfileOut( __METHOD__ );
1785                 return $retVal;
1786         }
1787
1788         /**
1789          * Render a forced-blue link inline; protect against double expansion of
1790          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1791          * Since this little disaster has to split off the trail text to avoid
1792          * breaking URLs in the following text without breaking trails on the
1793          * wiki links, it's been made into a horrible function.
1794          *
1795          * @param Title $nt
1796          * @param string $text
1797          * @param string $query
1798          * @param string $trail
1799          * @param string $prefix
1800          * @return string HTML-wikitext mix oh yuck
1801          */
1802         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1803                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1804                 $sk = $this->mOptions->getSkin();
1805                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1806                 return $this->armorLinks( $link ) . $trail;
1807         }
1808
1809         /**
1810          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1811          * going to go through further parsing steps before inline URL expansion.
1812          *
1813          * In particular this is important when using action=render, which causes
1814          * full URLs to be included.
1815          *
1816          * Oh man I hate our multi-layer parser!
1817          *
1818          * @param string more-or-less HTML
1819          * @return string less-or-more HTML with NOPARSE bits
1820          */
1821         function armorLinks( $text ) {
1822                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1823                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1824         }
1825
1826         /**
1827          * Return true if subpage links should be expanded on this page.
1828          * @return bool
1829          */
1830         function areSubpagesAllowed() {
1831                 # Some namespaces don't allow subpages
1832                 global $wgNamespacesWithSubpages;
1833                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1834         }
1835
1836         /**
1837          * Handle link to subpage if necessary
1838          * @param string $target the source of the link
1839          * @param string &$text the link text, modified as necessary
1840          * @return string the full name of the link
1841          * @private
1842          */
1843         function maybeDoSubpageLink($target, &$text) {
1844                 # Valid link forms:
1845                 # Foobar -- normal
1846                 # :Foobar -- override special treatment of prefix (images, language links)
1847                 # /Foobar -- convert to CurrentPage/Foobar
1848                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1849                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1850                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1851
1852                 $fname = 'Parser::maybeDoSubpageLink';
1853                 wfProfileIn( $fname );
1854                 $ret = $target; # default return value is no change
1855
1856                 # Some namespaces don't allow subpages,
1857                 # so only perform processing if subpages are allowed
1858                 if( $this->areSubpagesAllowed() ) {
1859                         $hash = strpos( $target, '#' );
1860                         if( $hash !== false ) {
1861                                 $suffix = substr( $target, $hash );
1862                                 $target = substr( $target, 0, $hash );
1863                         } else {
1864                                 $suffix = '';
1865                         }
1866                         # bug 7425
1867                         $target = trim( $target );
1868                         # Look at the first character
1869                         if( $target != '' && $target{0} == '/' ) {
1870                                 # / at end means we don't want the slash to be shown
1871                                 $m = array();
1872                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1873                                 if( $trailingSlashes ) {
1874                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1875                                 } else {
1876                                         $noslash = substr( $target, 1 );
1877                                 }
1878
1879                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1880                                 if( '' === $text ) {
1881                                         $text = $target . $suffix;
1882                                 } # this might be changed for ugliness reasons
1883                         } else {
1884                                 # check for .. subpage backlinks
1885                                 $dotdotcount = 0;
1886                                 $nodotdot = $target;
1887                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1888                                         ++$dotdotcount;
1889                                         $nodotdot = substr( $nodotdot, 3 );
1890                                 }
1891                                 if($dotdotcount > 0) {
1892                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1893                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1894                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1895                                                 # / at the end means don't show full path
1896                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1897                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1898                                                         if( '' === $text ) {
1899                                                                 $text = $nodotdot . $suffix;
1900                                                         }
1901                                                 }
1902                                                 $nodotdot = trim( $nodotdot );
1903                                                 if( $nodotdot != '' ) {
1904                                                         $ret .= '/' . $nodotdot;
1905                                                 }
1906                                                 $ret .= $suffix;
1907                                         }
1908                                 }
1909                         }
1910                 }
1911
1912                 wfProfileOut( $fname );
1913                 return $ret;
1914         }
1915
1916         /**#@+
1917          * Used by doBlockLevels()
1918          * @private
1919          */
1920         /* private */ function closeParagraph() {
1921                 $result = '';
1922                 if ( '' != $this->mLastSection ) {
1923                         $result = '</' . $this->mLastSection  . ">\n";
1924                 }
1925                 $this->mInPre = false;
1926                 $this->mLastSection = '';
1927                 return $result;
1928         }
1929         # getCommon() returns the length of the longest common substring
1930         # of both arguments, starting at the beginning of both.
1931         #
1932         /* private */ function getCommon( $st1, $st2 ) {
1933                 $fl = strlen( $st1 );
1934                 $shorter = strlen( $st2 );
1935                 if ( $fl < $shorter ) { $shorter = $fl; }
1936
1937                 for ( $i = 0; $i < $shorter; ++$i ) {
1938                         if ( $st1{$i} != $st2{$i} ) { break; }
1939                 }
1940                 return $i;
1941         }
1942         # These next three functions open, continue, and close the list
1943         # element appropriate to the prefix character passed into them.
1944         #
1945         /* private */ function openList( $char ) {
1946                 $result = $this->closeParagraph();
1947
1948                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1949                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1950                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1951                 else if ( ';' == $char ) {
1952                         $result .= '<dl><dt>';
1953                         $this->mDTopen = true;
1954                 }
1955                 else { $result = '<!-- ERR 1 -->'; }
1956
1957                 return $result;
1958         }
1959
1960         /* private */ function nextItem( $char ) {
1961                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1962                 else if ( ':' == $char || ';' == $char ) {
1963                         $close = '</dd>';
1964                         if ( $this->mDTopen ) { $close = '</dt>'; }
1965                         if ( ';' == $char ) {
1966                                 $this->mDTopen = true;
1967                                 return $close . '<dt>';
1968                         } else {
1969                                 $this->mDTopen = false;
1970                                 return $close . '<dd>';
1971                         }
1972                 }
1973                 return '<!-- ERR 2 -->';
1974         }
1975
1976         /* private */ function closeList( $char ) {
1977                 if ( '*' == $char ) { $text = '</li></ul>'; }
1978                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1979                 else if ( ':' == $char ) {
1980                         if ( $this->mDTopen ) {
1981                                 $this->mDTopen = false;
1982                                 $text = '</dt></dl>';
1983                         } else {
1984                                 $text = '</dd></dl>';
1985                         }
1986                 }
1987                 else {  return '<!-- ERR 3 -->'; }
1988                 return $text."\n";
1989         }
1990         /**#@-*/
1991
1992         /**
1993          * Make lists from lines starting with ':', '*', '#', etc.
1994          *
1995          * @private
1996          * @return string the lists rendered as HTML
1997          */
1998         function doBlockLevels( $text, $linestart ) {
1999                 $fname = 'Parser::doBlockLevels';
2000                 wfProfileIn( $fname );
2001
2002                 # Parsing through the text line by line.  The main thing
2003                 # happening here is handling of block-level elements p, pre,
2004                 # and making lists from lines starting with * # : etc.
2005                 #
2006                 $textLines = explode( "\n", $text );
2007
2008                 $lastPrefix = $output = '';
2009                 $this->mDTopen = $inBlockElem = false;
2010                 $prefixLength = 0;
2011                 $paragraphStack = false;
2012
2013                 if ( !$linestart ) {
2014                         $output .= array_shift( $textLines );
2015                 }
2016                 foreach ( $textLines as $oLine ) {
2017                         $lastPrefixLength = strlen( $lastPrefix );
2018                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2019                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2020                         if ( !$this->mInPre ) {
2021                                 # Multiple prefixes may abut each other for nested lists.
2022                                 $prefixLength = strspn( $oLine, '*#:;' );
2023                                 $pref = substr( $oLine, 0, $prefixLength );
2024
2025                                 # eh?
2026                                 $pref2 = str_replace( ';', ':', $pref );
2027                                 $t = substr( $oLine, $prefixLength );
2028                                 $this->mInPre = !empty($preOpenMatch);
2029                         } else {
2030                                 # Don't interpret any other prefixes in preformatted text
2031                                 $prefixLength = 0;
2032                                 $pref = $pref2 = '';
2033                                 $t = $oLine;
2034                         }
2035
2036                         # List generation
2037                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2038                                 # Same as the last item, so no need to deal with nesting or opening stuff
2039                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2040                                 $paragraphStack = false;
2041
2042                                 if ( substr( $pref, -1 ) == ';') {
2043                                         # The one nasty exception: definition lists work like this:
2044                                         # ; title : definition text
2045                                         # So we check for : in the remainder text to split up the
2046                                         # title and definition, without b0rking links.
2047                                         $term = $t2 = '';
2048                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2049                                                 $t = $t2;
2050                                                 $output .= $term . $this->nextItem( ':' );
2051                                         }
2052                                 }
2053                         } elseif( $prefixLength || $lastPrefixLength ) {
2054                                 # Either open or close a level...
2055                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2056                                 $paragraphStack = false;
2057
2058                                 while( $commonPrefixLength < $lastPrefixLength ) {
2059                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2060                                         --$lastPrefixLength;
2061                                 }
2062                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2063                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2064                                 }
2065                                 while ( $prefixLength > $commonPrefixLength ) {
2066                                         $char = substr( $pref, $commonPrefixLength, 1 );
2067                                         $output .= $this->openList( $char );
2068
2069                                         if ( ';' == $char ) {
2070                                                 # FIXME: This is dupe of code above
2071                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2072                                                         $t = $t2;
2073                                                         $output .= $term . $this->nextItem( ':' );
2074                                                 }
2075                                         }
2076                                         ++$commonPrefixLength;
2077                                 }
2078                                 $lastPrefix = $pref2;
2079                         }
2080                         if( 0 == $prefixLength ) {
2081                                 wfProfileIn( "$fname-paragraph" );
2082                                 # No prefix (not in list)--go to paragraph mode
2083                                 // XXX: use a stack for nestable elements like span, table and div
2084                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2085                                 $closematch = preg_match(
2086                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2087                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2088                                 if ( $openmatch or $closematch ) {
2089                                         $paragraphStack = false;
2090                                         # TODO bug 5718: paragraph closed
2091                                         $output .= $this->closeParagraph();
2092                                         if ( $preOpenMatch and !$preCloseMatch ) {
2093                                                 $this->mInPre = true;
2094                                         }
2095                                         if ( $closematch ) {
2096                                                 $inBlockElem = false;
2097                                         } else {
2098                                                 $inBlockElem = true;
2099                                         }
2100                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2101                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2102                                                 // pre
2103                                                 if ($this->mLastSection != 'pre') {
2104                                                         $paragraphStack = false;
2105                                                         $output .= $this->closeParagraph().'<pre>';
2106                                                         $this->mLastSection = 'pre';
2107                                                 }
2108                                                 $t = substr( $t, 1 );
2109                                         } else {
2110                                                 // paragraph
2111                                                 if ( '' == trim($t) ) {
2112                                                         if ( $paragraphStack ) {
2113                                                                 $output .= $paragraphStack.'<br />';
2114                                                                 $paragraphStack = false;
2115                                                                 $this->mLastSection = 'p';
2116                                                         } else {
2117                                                                 if ($this->mLastSection != 'p' ) {
2118                                                                         $output .= $this->closeParagraph();
2119                                                                         $this->mLastSection = '';
2120                                                                         $paragraphStack = '<p>';
2121                                                                 } else {
2122                                                                         $paragraphStack = '</p><p>';
2123                                                                 }
2124                                                         }
2125                                                 } else {
2126                                                         if ( $paragraphStack ) {
2127                                                                 $output .= $paragraphStack;
2128                                                                 $paragraphStack = false;
2129                                                                 $this->mLastSection = 'p';
2130                                                         } else if ($this->mLastSection != 'p') {
2131                                                                 $output .= $this->closeParagraph().'<p>';
2132                                                                 $this->mLastSection = 'p';
2133                                                         }
2134                                                 }
2135                                         }
2136                                 }
2137                                 wfProfileOut( "$fname-paragraph" );
2138                         }
2139                         // somewhere above we forget to get out of pre block (bug 785)
2140                         if($preCloseMatch && $this->mInPre) {
2141                                 $this->mInPre = false;
2142                         }
2143                         if ($paragraphStack === false) {
2144                                 $output .= $t."\n";
2145                         }
2146                 }
2147                 while ( $prefixLength ) {
2148                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2149                         --$prefixLength;
2150                 }
2151                 if ( '' != $this->mLastSection ) {
2152                         $output .= '</' . $this->mLastSection . '>';
2153                         $this->mLastSection = '';
2154                 }
2155
2156                 wfProfileOut( $fname );
2157                 return $output;
2158         }
2159
2160         /**
2161          * Split up a string on ':', ignoring any occurences inside tags
2162          * to prevent illegal overlapping.
2163          * @param string $str the string to split
2164          * @param string &$before set to everything before the ':'
2165          * @param string &$after set to everything after the ':'
2166          * return string the position of the ':', or false if none found
2167          */
2168         function findColonNoLinks($str, &$before, &$after) {
2169                 $fname = 'Parser::findColonNoLinks';
2170                 wfProfileIn( $fname );
2171
2172                 $pos = strpos( $str, ':' );
2173                 if( $pos === false ) {
2174                         // Nothing to find!
2175                         wfProfileOut( $fname );
2176                         return false;
2177                 }
2178
2179                 $lt = strpos( $str, '<' );
2180                 if( $lt === false || $lt > $pos ) {
2181                         // Easy; no tag nesting to worry about
2182                         $before = substr( $str, 0, $pos );
2183                         $after = substr( $str, $pos+1 );
2184                         wfProfileOut( $fname );
2185                         return $pos;
2186                 }
2187
2188                 // Ugly state machine to walk through avoiding tags.
2189                 $state = self::COLON_STATE_TEXT;
2190                 $stack = 0;
2191                 $len = strlen( $str );
2192                 for( $i = 0; $i < $len; $i++ ) {
2193                         $c = $str{$i};
2194
2195                         switch( $state ) {
2196                         // (Using the number is a performance hack for common cases)
2197                         case 0: // self::COLON_STATE_TEXT:
2198                                 switch( $c ) {
2199                                 case "<":
2200                                         // Could be either a <start> tag or an </end> tag
2201                                         $state = self::COLON_STATE_TAGSTART;
2202                                         break;
2203                                 case ":":
2204                                         if( $stack == 0 ) {
2205                                                 // We found it!
2206                                                 $before = substr( $str, 0, $i );
2207                                                 $after = substr( $str, $i + 1 );
2208                                                 wfProfileOut( $fname );
2209                                                 return $i;
2210                                         }
2211                                         // Embedded in a tag; don't break it.
2212                                         break;
2213                                 default:
2214                                         // Skip ahead looking for something interesting
2215                                         $colon = strpos( $str, ':', $i );
2216                                         if( $colon === false ) {
2217                                                 // Nothing else interesting
2218                                                 wfProfileOut( $fname );
2219                                                 return false;
2220                                         }
2221                                         $lt = strpos( $str, '<', $i );
2222                                         if( $stack === 0 ) {
2223                                                 if( $lt === false || $colon < $lt ) {
2224                                                         // We found it!
2225                                                         $before = substr( $str, 0, $colon );
2226                                                         $after = substr( $str, $colon + 1 );
2227                                                         wfProfileOut( $fname );
2228                                                         return $i;
2229                                                 }
2230                                         }
2231                                         if( $lt === false ) {
2232                                                 // Nothing else interesting to find; abort!
2233                                                 // We're nested, but there's no close tags left. Abort!
2234                                                 break 2;
2235                                         }
2236                                         // Skip ahead to next tag start
2237                                         $i = $lt;
2238                                         $state = self::COLON_STATE_TAGSTART;
2239                                 }
2240                                 break;
2241                         case 1: // self::COLON_STATE_TAG:
2242                                 // In a <tag>
2243                                 switch( $c ) {
2244                                 case ">":
2245                                         $stack++;
2246                                         $state = self::COLON_STATE_TEXT;
2247                                         break;
2248                                 case "/":
2249                                         // Slash may be followed by >?
2250                                         $state = self::COLON_STATE_TAGSLASH;
2251                                         break;
2252                                 default:
2253                                         // ignore
2254                                 }
2255                                 break;
2256                         case 2: // self::COLON_STATE_TAGSTART:
2257                                 switch( $c ) {
2258                                 case "/":
2259                                         $state = self::COLON_STATE_CLOSETAG;
2260                                         break;
2261                                 case "!":
2262                                         $state = self::COLON_STATE_COMMENT;
2263                                         break;
2264                                 case ">":
2265                                         // Illegal early close? This shouldn't happen D:
2266                                         $state = self::COLON_STATE_TEXT;
2267                                         break;
2268                                 default:
2269                                         $state = self::COLON_STATE_TAG;
2270                                 }
2271                                 break;
2272                         case 3: // self::COLON_STATE_CLOSETAG:
2273                                 // In a </tag>
2274                                 if( $c == ">" ) {
2275                                         $stack--;
2276                                         if( $stack < 0 ) {
2277                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2278                                                 wfProfileOut( $fname );
2279                                                 return false;
2280                                         }
2281                                         $state = self::COLON_STATE_TEXT;
2282                                 }
2283                                 break;
2284                         case self::COLON_STATE_TAGSLASH:
2285                                 if( $c == ">" ) {
2286                                         // Yes, a self-closed tag <blah/>
2287                                         $state = self::COLON_STATE_TEXT;
2288                                 } else {
2289                                         // Probably we're jumping the gun, and this is an attribute
2290                                         $state = self::COLON_STATE_TAG;
2291                                 }
2292                                 break;
2293                         case 5: // self::COLON_STATE_COMMENT:
2294                                 if( $c == "-" ) {
2295                                         $state = self::COLON_STATE_COMMENTDASH;
2296                                 }
2297                                 break;
2298                         case self::COLON_STATE_COMMENTDASH:
2299                                 if( $c == "-" ) {
2300                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2301                                 } else {
2302                                         $state = self::COLON_STATE_COMMENT;
2303                                 }
2304                                 break;
2305                         case self::COLON_STATE_COMMENTDASHDASH:
2306                                 if( $c == ">" ) {
2307                                         $state = self::COLON_STATE_TEXT;
2308                                 } else {
2309                                         $state = self::COLON_STATE_COMMENT;
2310                                 }
2311                                 break;
2312                         default:
2313                                 throw new MWException( "State machine error in $fname" );
2314                         }
2315                 }
2316                 if( $stack > 0 ) {
2317                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2318                         return false;
2319                 }
2320                 wfProfileOut( $fname );
2321                 return false;
2322         }
2323
2324         /**
2325          * Return value of a magic variable (like PAGENAME)
2326          *
2327          * @private
2328          */
2329         function getVariableValue( $index ) {
2330                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2331
2332                 /**
2333                  * Some of these require message or data lookups and can be
2334                  * expensive to check many times.
2335                  */
2336                 static $varCache = array();
2337                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2338                         if ( isset( $varCache[$index] ) ) {
2339                                 return $varCache[$index];
2340                         }
2341                 }
2342
2343                 $ts = time();
2344                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2345
2346                 # Use the time zone
2347                 global $wgLocaltimezone;
2348                 if ( isset( $wgLocaltimezone ) ) {
2349                         $oldtz = getenv( 'TZ' );
2350                         putenv( 'TZ='.$wgLocaltimezone );
2351                 }
2352
2353                 wfSuppressWarnings(); // E_STRICT system time bitching
2354                 $localTimestamp = date( 'YmdHis', $ts );
2355                 $localMonth = date( 'm', $ts );
2356                 $localMonthName = date( 'n', $ts );
2357                 $localDay = date( 'j', $ts );
2358                 $localDay2 = date( 'd', $ts );
2359                 $localDayOfWeek = date( 'w', $ts );
2360                 $localWeek = date( 'W', $ts );
2361                 $localYear = date( 'Y', $ts );
2362                 $localHour = date( 'H', $ts );
2363                 if ( isset( $wgLocaltimezone ) ) {
2364                         putenv( 'TZ='.$oldtz );
2365                 }
2366                 wfRestoreWarnings();
2367
2368                 switch ( $index ) {
2369                         case 'currentmonth':
2370                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2371                         case 'currentmonthname':
2372                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2373                         case 'currentmonthnamegen':
2374                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2375                         case 'currentmonthabbrev':
2376                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2377                         case 'currentday':
2378                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2379                         case 'currentday2':
2380                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2381                         case 'localmonth':
2382                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2383                         case 'localmonthname':
2384                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2385                         case 'localmonthnamegen':
2386                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2387                         case 'localmonthabbrev':
2388                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2389                         case 'localday':
2390                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2391                         case 'localday2':
2392                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2393                         case 'pagename':
2394                                 return wfEscapeWikiText( $this->mTitle->getText() );
2395                         case 'pagenamee':
2396                                 return $this->mTitle->getPartialURL();
2397                         case 'fullpagename':
2398                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2399                         case 'fullpagenamee':
2400                                 return $this->mTitle->getPrefixedURL();
2401                         case 'subpagename':
2402                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2403                         case 'subpagenamee':
2404                                 return $this->mTitle->getSubpageUrlForm();
2405                         case 'basepagename':
2406                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2407                         case 'basepagenamee':
2408                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2409                         case 'talkpagename':
2410                                 if( $this->mTitle->canTalk() ) {
2411                                         $talkPage = $this->mTitle->getTalkPage();
2412                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2413                                 } else {
2414                                         return '';
2415                                 }
2416                         case 'talkpagenamee':
2417                                 if( $this->mTitle->canTalk() ) {
2418                                         $talkPage = $this->mTitle->getTalkPage();
2419                                         return $talkPage->getPrefixedUrl();
2420                                 } else {
2421                                         return '';
2422                                 }
2423                         case 'subjectpagename':
2424                                 $subjPage = $this->mTitle->getSubjectPage();
2425                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2426                         case 'subjectpagenamee':
2427                                 $subjPage = $this->mTitle->getSubjectPage();
2428                                 return $subjPage->getPrefixedUrl();
2429                         case 'revisionid':
2430                                 // Let the edit saving system know we should parse the page
2431                                 // *after* a revision ID has been assigned.
2432                                 $this->mOutput->setFlag( 'vary-revision' );
2433                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2434                                 return $this->mRevisionId;
2435                         case 'revisionday':
2436                                 // Let the edit saving system know we should parse the page
2437                                 // *after* a revision ID has been assigned. This is for null edits.
2438                                 $this->mOutput->setFlag( 'vary-revision' );
2439                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2440                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2441                         case 'revisionday2':
2442                                 // Let the edit saving system know we should parse the page
2443                                 // *after* a revision ID has been assigned. This is for null edits.
2444                                 $this->mOutput->setFlag( 'vary-revision' );
2445                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2446                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2447                         case 'revisionmonth':
2448                                 // Let the edit saving system know we should parse the page
2449                                 // *after* a revision ID has been assigned. This is for null edits.
2450                                 $this->mOutput->setFlag( 'vary-revision' );
2451                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2452                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2453                         case 'revisionyear':
2454                                 // Let the edit saving system know we should parse the page
2455                                 // *after* a revision ID has been assigned. This is for null edits.
2456                                 $this->mOutput->setFlag( 'vary-revision' );
2457                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2458                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2459                         case 'revisiontimestamp':
2460                                 // Let the edit saving system know we should parse the page
2461                                 // *after* a revision ID has been assigned. This is for null edits.
2462                                 $this->mOutput->setFlag( 'vary-revision' );
2463                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2464                                 return $this->getRevisionTimestamp();
2465                         case 'namespace':
2466                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2467                         case 'namespacee':
2468                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2469                         case 'talkspace':
2470                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2471                         case 'talkspacee':
2472                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2473                         case 'subjectspace':
2474                                 return $this->mTitle->getSubjectNsText();
2475                         case 'subjectspacee':
2476                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2477                         case 'currentdayname':
2478                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2479                         case 'currentyear':
2480                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2481                         case 'currenttime':
2482                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2483                         case 'currenthour':
2484                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2485                         case 'currentweek':
2486                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2487                                 // int to remove the padding
2488                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2489                         case 'currentdow':
2490                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2491                         case 'localdayname':
2492                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2493                         case 'localyear':
2494                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2495                         case 'localtime':
2496                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2497                         case 'localhour':
2498                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2499                         case 'localweek':
2500                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2501                                 // int to remove the padding
2502                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2503                         case 'localdow':
2504                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2505                         case 'numberofarticles':
2506                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2507                         case 'numberoffiles':
2508                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2509                         case 'numberofusers':
2510                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2511                         case 'numberofpages':
2512                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2513                         case 'numberofadmins':
2514                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2515                         case 'numberofedits':
2516                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2517                         case 'currenttimestamp':
2518                                 return $varCache[$index] = wfTimestampNow();
2519                         case 'localtimestamp':
2520                                 return $varCache[$index] = $localTimestamp;
2521                         case 'currentversion':
2522                                 return $varCache[$index] = SpecialVersion::getVersion();
2523                         case 'sitename':
2524                                 return $wgSitename;
2525                         case 'server':
2526                                 return $wgServer;
2527                         case 'servername':
2528                                 return $wgServerName;
2529                         case 'scriptpath':
2530                                 return $wgScriptPath;
2531                         case 'directionmark':
2532                                 return $wgContLang->getDirMark();
2533                         case 'contentlanguage':
2534                                 global $wgContLanguageCode;
2535                                 return $wgContLanguageCode;
2536                         default:
2537                                 $ret = null;
2538                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2539                                         return $ret;
2540                                 else
2541                                         return null;
2542                 }
2543         }
2544
2545         /**
2546          * initialise the magic variables (like CURRENTMONTHNAME)
2547          *
2548          * @private
2549          */
2550         function initialiseVariables() {
2551                 $fname = 'Parser::initialiseVariables';
2552                 wfProfileIn( $fname );
2553                 $variableIDs = MagicWord::getVariableIDs();
2554
2555                 $this->mVariables = new MagicWordArray( $variableIDs );
2556                 wfProfileOut( $fname );
2557         }
2558
2559         /**
2560          * Preprocess some wikitext and return the document tree.
2561          * This is the ghost of replace_variables().
2562          *
2563          * @param string $text The text to parse
2564          * @param integer flags Bitwise combination of:
2565          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2566          *                                     included. Default is to assume a direct page view.
2567          *
2568          * The generated DOM tree must depend only on the input text, the flags, and $this->ot['msg'].
2569          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2570          *
2571          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2572          * change in the DOM tree for a given text, must be passed through the section identifier
2573          * in the section edit link and thus back to extractSections().
2574          *
2575          * The output of this function is currently only cached in process memory, but a persistent
2576          * cache may be implemented at a later date which takes further advantage of these strict
2577          * dependency requirements.
2578          *
2579          * @private
2580          */
2581         function preprocessToDom ( $text, $flags = 0 ) {
2582                 wfProfileIn( __METHOD__ );
2583                 wfProfileIn( __METHOD__.'-makexml' );
2584
2585                 static $msgRules, $normalRules, $inclusionSupertags, $nonInclusionSupertags;
2586                 if ( !$msgRules ) {
2587                         $msgRules = array(
2588                                 '{' => array(
2589                                         'end' => '}',
2590                                         'names' => array(
2591                                                 2 => 'template',
2592                                         ),
2593                                         'min' => 2,
2594                                         'max' => 2,
2595                                 ),
2596                                 '[' => array(
2597                                         'end' => ']',
2598                                         'names' => array( 2 => null ),
2599                                         'min' => 2,
2600                                         'max' => 2,
2601                                 )
2602                         );
2603                         $normalRules = array(
2604                                 '{' => array(
2605                                         'end' => '}',
2606                                         'names' => array(
2607                                                 2 => 'template',
2608                                                 3 => 'tplarg',
2609                                         ),
2610                                         'min' => 2,
2611                                         'max' => 3,
2612                                 ),
2613                                 '[' => array(
2614                                         'end' => ']',
2615                                         'names' => array( 2 => null ),
2616                                         'min' => 2,
2617                                         'max' => 2,
2618                                 )
2619                         );
2620                 }
2621                 if ( $this->ot['msg'] ) {
2622                         $rules = $msgRules;
2623                 } else {
2624                         $rules = $normalRules;
2625                 }
2626                 $forInclusion = $flags & self::PTD_FOR_INCLUSION;
2627
2628                 $xmlishElements = $this->getStripList();
2629                 $enableOnlyinclude = false;
2630                 if ( $forInclusion ) {
2631                         $ignoredTags = array( 'includeonly', '/includeonly' );
2632                         $ignoredElements = array( 'noinclude' );
2633                         $xmlishElements[] = 'noinclude';
2634                         if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
2635                                 $enableOnlyinclude = true;
2636                         }
2637                 } else {
2638                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
2639                         $ignoredElements = array( 'includeonly' );
2640                         $xmlishElements[] = 'includeonly';
2641                 }
2642                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
2643
2644                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2645                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
2646
2647                 $stack = array();      # Stack of unclosed parentheses
2648                 $stackIndex = -1;      # Stack read pointer
2649
2650                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2651                 $revText = strrev( $text ); // For fast reverse searches
2652
2653                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2654                 $topAccum = '<root>';      # Top level text accumulator
2655                 $accum =& $topAccum;            # Current text accumulator
2656                 $findEquals = false;            # True to find equals signs in arguments
2657                 $findHeading = false;           # True to look at LF characters for possible headings
2658                 $findPipe = false;              # True to take notice of pipe characters
2659                 $headingIndex = 1;
2660                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2661                 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
2662
2663                 if ( $enableOnlyinclude ) {
2664                         $i = 0;
2665                 }
2666
2667                 while ( true ) {
2668                         if ( $findOnlyinclude ) {
2669                                 // Ignore all input up to the next <onlyinclude>
2670                                 $startPos = strpos( $text, '<onlyinclude>', $i );
2671                                 if ( $startPos === false ) {
2672                                         // Ignored section runs to the end
2673                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
2674                                         break;
2675                                 }
2676                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
2677                                 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
2678                                 $i = $tagEndPos;
2679                                 $findOnlyinclude = false;
2680                         }
2681
2682                         if ( $i == -1 ) {
2683                                 $found = 'line-start';
2684                                 $curChar = '';
2685                         } else {
2686                                 # Find next opening brace, closing brace or pipe
2687                                 $search = $searchBase;
2688                                 if ( $stackIndex == -1 ) {
2689                                         $currentClosing = '';
2690                                         // Look for headings only at the top stack level
2691                                         // Among other things, this resolves the ambiguity between =
2692                                         // for headings and = for template arguments
2693                                         $search .= "\n";
2694                                 } else {
2695                                         $currentClosing = $stack[$stackIndex]['close'];
2696                                         $search .= $currentClosing;
2697                                 }
2698                                 if ( $findPipe ) {
2699                                         $search .= '|';
2700                                 }
2701                                 if ( $findEquals ) {
2702                                         $search .= '=';
2703                                 }
2704                                 $rule = null;
2705                                 # Output literal section, advance input counter
2706                                 $literalLength = strcspn( $text, $search, $i );
2707                                 if ( $literalLength > 0 ) {
2708                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2709                                         $i += $literalLength;
2710                                 }
2711                                 if ( $i >= strlen( $text ) ) {
2712                                         if ( $currentClosing == "\n" ) {
2713                                                 // Do a past-the-end run to finish off the heading
2714                                                 $curChar = '';
2715                                                 $found = 'line-end';
2716                                         } else {
2717                                                 # All done
2718                                                 break;
2719                                         }
2720                                 } else {
2721                                         $curChar = $text[$i];
2722                                         if ( $curChar == '|' ) {
2723                                                 $found = 'pipe';
2724                                         } elseif ( $curChar == '=' ) {
2725                                                 $found = 'equals';
2726                                         } elseif ( $curChar == '<' ) {
2727                                                 $found = 'angle';
2728                                         } elseif ( $curChar == "\n" ) {
2729                                                 if ( $stackIndex == -1 ) {
2730                                                         $found = 'line-start';
2731                                                 } else {
2732                                                         $found = 'line-end';
2733                                                 }
2734                                         } elseif ( $curChar == $currentClosing ) {
2735                                                 $found = 'close';
2736                                         } elseif ( isset( $rules[$curChar] ) ) {
2737                                                 $found = 'open';
2738                                                 $rule = $rules[$curChar];
2739                                         } else {
2740                                                 # Some versions of PHP have a strcspn which stops on null characters
2741                                                 # Ignore and continue
2742                                                 ++$i;
2743                                                 continue;
2744                                         }
2745                                 }
2746                         }
2747
2748                         if ( $found == 'angle' ) {
2749                                 $matches = false;
2750                                 // Handle </onlyinclude>
2751                                 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
2752                                         $findOnlyinclude = true;
2753                                         continue;
2754                                 }
2755
2756                                 // Determine element name
2757                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
2758                                         // Element name missing or not listed
2759                                         $accum .= '&lt;';
2760                                         ++$i;
2761                                         continue;
2762                                 }
2763                                 // Handle comments
2764                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2765                                         // To avoid leaving blank lines, when a comment is both preceded
2766                                         // and followed by a newline (ignoring spaces), trim leading and
2767                                         // trailing spaces and one of the newlines.
2768
2769                                         // Find the end
2770                                         $endPos = strpos( $text, '-->', $i + 4 );
2771                                         if ( $endPos === false ) {
2772                                                 // Unclosed comment in input, runs to end
2773                                                 $inner = substr( $text, $i );
2774                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2775                                                 $i = strlen( $text );
2776                                         } else {
2777                                                 // Search backwards for leading whitespace
2778                                                 $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
2779                                                 // Search forwards for trailing whitespace
2780                                                 // $wsEnd will be the position of the last space
2781                                                 $wsEnd = $endPos + 2 + strspn( $text, ' ', $endPos + 3 );
2782                                                 // Eat the line if possible
2783                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
2784                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n" )
2785                                                 {
2786                                                         $startPos = $wsStart;
2787                                                         $endPos = $wsEnd + 1;
2788                                                         // Remove leading whitespace from the end of the accumulator
2789                                                         // Sanity check first though
2790                                                         $wsLength = $i - $wsStart;
2791                                                         if ( $wsLength > 0 && substr( $accum, -$wsLength ) === str_repeat( ' ', $wsLength ) ) {
2792                                                                 $accum = substr( $accum, 0, -$wsLength );
2793                                                         }
2794                                                 } else {
2795                                                         // No line to eat, just take the comment itself
2796                                                         $startPos = $i;
2797                                                         $endPos += 2;
2798                                                 }
2799
2800                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
2801                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2802                                                 $i = $endPos + 1;
2803                                         }
2804                                         continue;
2805                                 }
2806                                 $name = $matches[1];
2807                                 $attrStart = $i + strlen( $name ) + 1;
2808
2809                                 // Find end of tag
2810                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2811                                 if ( $tagEndPos === false ) {
2812                                         // Infinite backtrack
2813                                         // Disable tag search to prevent worst-case O(N^2) performance
2814                                         $noMoreGT = true;
2815                                         $accum .= '&lt;';
2816                                         ++$i;
2817                                         continue;
2818                                 }
2819
2820                                 // Handle ignored tags
2821                                 if ( in_array( $name, $ignoredTags ) ) {
2822                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '</ignore>';
2823                                         $i = $tagEndPos + 1;
2824                                         continue;
2825                                 }
2826
2827                                 $tagStartPos = $i;
2828                                 if ( $text[$tagEndPos-1] == '/' ) {
2829                                         $attrEnd = $tagEndPos - 1;
2830                                         $inner = null;
2831                                         $i = $tagEndPos + 1;
2832                                         $close = '';
2833                                 } else {
2834                                         $attrEnd = $tagEndPos;
2835                                         // Find closing tag
2836                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2837                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2838                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2839                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2840                                         } else {
2841                                                 // No end tag -- let it run out to the end of the text.
2842                                                 $inner = substr( $text, $tagEndPos + 1 );
2843                                                 $i = strlen( $text );
2844                                                 $close = '';
2845                                         }
2846                                 }
2847                                 // <includeonly> and <noinclude> just become <ignore> tags
2848                                 if ( in_array( $name, $ignoredElements ) ) {
2849                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
2850                                                 . '</ignore>';
2851                                         continue;
2852                                 }
2853
2854                                 $accum .= '<ext>';
2855                                 if ( $attrEnd <= $attrStart ) {
2856                                         $attr = '';
2857                                 } else {
2858                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2859                                 }
2860                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2861                                         // Note that the attr element contains the whitespace between name and attribute,
2862                                         // this is necessary for precise reconstruction during pre-save transform.
2863                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2864                                 if ( $inner !== null ) {
2865                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2866                                 }
2867                                 $accum .= $close . '</ext>';
2868                         }
2869
2870                         elseif ( $found == 'line-start' ) {
2871                                 // Is this the start of a heading?
2872                                 // Line break belongs before the heading element in any case
2873                                 $accum .= $curChar;
2874                                 $i++;
2875
2876                                 $count = strspn( $text, '=', $i, 6 );
2877                                 if ( $count > 0 ) {
2878                                         $piece = array(
2879                                                 'open' => "\n",
2880                                                 'close' => "\n",
2881                                                 'parts' => array( str_repeat( '=', $count ) ),
2882                                                 'count' => $count );
2883                                         $stack[++$stackIndex] = $piece;
2884                                         $i += $count;
2885                                         $accum =& $stack[$stackIndex]['parts'][0];
2886                                         $findPipe = false;
2887                                 }
2888                         }
2889
2890                         elseif ( $found == 'line-end' ) {
2891                                 $piece = $stack[$stackIndex];
2892                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2893                                 assert( $piece['open'] == "\n" );
2894                                 assert( $stackIndex == 0 );
2895                                 // Search back through the input to see if it has a proper close
2896                                 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
2897                                 $m = false;
2898                                 $count = $piece['count'];
2899                                 if ( preg_match( "/\s*(={{$count}})/A", $revText, $m, 0, strlen( $text ) - $i ) ) {
2900                                         // Found match, output <h>
2901                                         $count = min( strlen( $m[1] ), $count );
2902                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2903                                         $headingIndex++;
2904                                 } else {
2905                                         // No match, no <h>, just pass down the inner text
2906                                         $element = $accum;
2907                                 }
2908                                 // Unwind the stack
2909                                 // Headings can only occur on the top level, so this is a bit simpler than the
2910                                 // generic stack unwind operation in the close case
2911                                 unset( $stack[$stackIndex--] );
2912                                 $accum =& $topAccum;
2913                                 $findEquals = false;
2914                                 $findPipe = false;
2915
2916                                 // Append the result to the enclosing accumulator
2917                                 $accum .= $element;
2918                                 // Note that we do NOT increment the input pointer.
2919                                 // This is because the closing linebreak could be the opening linebreak of
2920                                 // another heading. Infinite loops are avoided because the next iteration MUST
2921                                 // hit the heading open case above, which unconditionally increments the
2922                                 // input pointer.
2923                         }
2924
2925                         elseif ( $found == 'open' ) {
2926                                 # count opening brace characters
2927                                 $count = strspn( $text, $curChar, $i );
2928
2929                                 # we need to add to stack only if opening brace count is enough for one of the rules
2930                                 if ( $count >= $rule['min'] ) {
2931                                         # Add it to the stack
2932                                         $piece = array(
2933                                                 'open' => $curChar,
2934                                                 'close' => $rule['end'],
2935                                                 'count' => $count,
2936                                                 'parts' => array( '' ),
2937                                                 'eqpos' => array(),
2938                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2939                                         );
2940
2941                                         $stackIndex ++;
2942                                         $stack[$stackIndex] = $piece;
2943                                         $accum =& $stack[$stackIndex]['parts'][0];
2944                                         $findEquals = false;
2945                                         $findPipe = true;
2946                                 } else {
2947                                         # Add literal brace(s)
2948                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2949                                 }
2950                                 $i += $count;
2951                         }
2952
2953                         elseif ( $found == 'close' ) {
2954                                 $piece = $stack[$stackIndex];
2955                                 # lets check if there are enough characters for closing brace
2956                                 $maxCount = $piece['count'];
2957                                 $count = strspn( $text, $curChar, $i, $maxCount );
2958
2959                                 # check for maximum matching characters (if there are 5 closing
2960                                 # characters, we will probably need only 3 - depending on the rules)
2961                                 $matchingCount = 0;
2962                                 $rule = $rules[$piece['open']];
2963                                 if ( $count > $rule['max'] ) {
2964                                         # The specified maximum exists in the callback array, unless the caller
2965                                         # has made an error
2966                                         $matchingCount = $rule['max'];
2967                                 } else {
2968                                         # Count is less than the maximum
2969                                         # Skip any gaps in the callback array to find the true largest match
2970                                         # Need to use array_key_exists not isset because the callback can be null
2971                                         $matchingCount = $count;
2972                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2973                                                 --$matchingCount;
2974                                         }
2975                                 }
2976
2977                                 if ($matchingCount <= 0) {
2978                                         # No matching element found in callback array
2979                                         # Output a literal closing brace and continue
2980                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2981                                         $i += $count;
2982                                         continue;
2983                                 }
2984                                 $name = $rule['names'][$matchingCount];
2985                                 if ( $name === null ) {
2986                                         // No element, just literal text
2987                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2988                                                 implode( '|', $piece['parts'] ) .
2989                                                 str_repeat( $rule['end'], $matchingCount );
2990                                 } else {
2991                                         # Create XML element
2992                                         # Note: $parts is already XML, does not need to be encoded further
2993                                         $parts = $piece['parts'];
2994                                         $title = $parts[0];
2995                                         unset( $parts[0] );
2996
2997                                         # The invocation is at the start of the line if lineStart is set in
2998                                         # the stack, and all opening brackets are used up.
2999                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
3000                                                 $attr = ' lineStart="1"';
3001                                         } else {
3002                                                 $attr = '';
3003                                         }
3004
3005                                         $element = "<$name$attr>";
3006                                         $element .= "<title>$title</title>";
3007                                         $argIndex = 1;
3008                                         foreach ( $parts as $partIndex => $part ) {
3009                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
3010                                                         $eqpos = $piece['eqpos'][$partIndex];
3011                                                         $argName = substr( $part, 0, $eqpos );
3012                                                         $argValue = substr( $part, $eqpos + 1 );
3013                                                         $element .= "<part><name>$argName</name>=<value>$argValue</value></part>";
3014                                                 } else {
3015                                                         $element .= "<part><name index=\"$argIndex\" /><value>$part</value></part>";
3016                                                         $argIndex++;
3017                                                 }
3018                                         }
3019                                         $element .= "</$name>";
3020                                 }
3021
3022                                 # Advance input pointer
3023                                 $i += $matchingCount;
3024
3025                                 # Unwind the stack
3026                                 unset( $stack[$stackIndex--] );
3027                                 if ( $stackIndex == -1 ) {
3028                                         $accum =& $topAccum;
3029                                         $findEquals = false;
3030                                         $findPipe = false;
3031                                 } else {
3032                                         $partCount = count( $stack[$stackIndex]['parts'] );
3033                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
3034                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
3035                                         $findEquals = $findPipe && $partCount > 1
3036                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
3037                                 }
3038
3039                                 # Re-add the old stack element if it still has unmatched opening characters remaining
3040                                 if ($matchingCount < $piece['count']) {
3041                                         $piece['parts'] = array( '' );
3042                                         $piece['count'] -= $matchingCount;
3043                                         $piece['eqpos'] = array();
3044                                         # do we still qualify for any callback with remaining count?
3045                                         $names = $rules[$piece['open']]['names'];
3046                                         $skippedBraces = 0;
3047                                         $enclosingAccum =& $accum;
3048                                         while ( $piece['count'] ) {
3049                                                 if ( array_key_exists( $piece['count'], $names ) ) {
3050                                                         $stackIndex++;
3051                                                         $stack[$stackIndex] = $piece;
3052                                                         $accum =& $stack[$stackIndex]['parts'][0];
3053                                                         $findEquals = true;
3054                                                         $findPipe = true;
3055                                                         break;
3056                                                 }
3057                                                 --$piece['count'];
3058                                                 $skippedBraces ++;
3059                                         }
3060                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
3061                                 }
3062
3063                                 # Add XML element to the enclosing accumulator
3064                                 $accum .= $element;
3065                         }
3066
3067                         elseif ( $found == 'pipe' ) {
3068                                 $stack[$stackIndex]['parts'][] = '';
3069                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3070                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
3071                                 $findEquals = true;
3072                                 ++$i;
3073                         }
3074
3075                         elseif ( $found == 'equals' ) {
3076                                 $findEquals = false;
3077                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3078                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
3079                                 $accum .= '=';
3080                                 ++$i;
3081                         }
3082                 }
3083
3084                 # Output any remaining unclosed brackets
3085                 foreach ( $stack as $piece ) {
3086                         if ( $piece['open'] == "\n" ) {
3087                                 $topAccum .= $piece['parts'][0];
3088                         } else {
3089                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
3090                         }
3091                 }
3092                 $topAccum .= '</root>';
3093
3094                 wfProfileOut( __METHOD__.'-makexml' );
3095                 wfProfileIn( __METHOD__.'-loadXML' );
3096                 $dom = new DOMDocument;
3097                 wfSuppressWarnings();
3098                 $result = $dom->loadXML( $topAccum );
3099                 wfRestoreWarnings();
3100                 if ( !$result ) {
3101                         // Try running the XML through UtfNormal to get rid of invalid characters
3102                         $topAccum = UtfNormal::cleanUp( $topAccum );
3103                         $result = $dom->loadXML( $topAccum );
3104                         if ( !$result ) {
3105                                 throw new MWException( __METHOD__.' generated invalid XML' );
3106                         }
3107                 }
3108                 wfProfileOut( __METHOD__.'-loadXML' );
3109                 wfProfileOut( __METHOD__ );
3110                 return $dom;
3111         }
3112
3113         /*
3114          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3115          */
3116         public static function splitWhitespace( $s ) {
3117                 $ltrimmed = ltrim( $s );
3118                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3119                 $trimmed = rtrim( $ltrimmed );
3120                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3121                 if ( $diff > 0 ) {
3122                         $w2 = substr( $ltrimmed, -$diff );
3123                 } else {
3124                         $w2 = '';
3125                 }
3126                 return array( $w1, $trimmed, $w2 );
3127         }
3128
3129         /**
3130          * Replace magic variables, templates, and template arguments
3131          * with the appropriate text. Templates are substituted recursively,
3132          * taking care to avoid infinite loops.
3133          *
3134          * Note that the substitution depends on value of $mOutputType:
3135          *  OT_WIKI: only {{subst:}} templates
3136          *  OT_MSG: only magic variables
3137          *  OT_HTML: all templates and magic variables
3138          *
3139          * @param string $tex The text to transform
3140          * @param PPFrame $frame Object describing the arguments passed to the template
3141          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3142          * @private
3143          */
3144         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3145                 # Prevent too big inclusions
3146                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3147                         return $text;
3148                 }
3149
3150                 $fname = __METHOD__;
3151                 wfProfileIn( $fname );
3152
3153                 if ( $frame === false ) {
3154                         $frame = new PPFrame( $this );
3155                 } elseif ( !( $frame instanceof PPFrame ) ) {
3156                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3157                 }
3158
3159                 $dom = $this->preprocessToDom( $text );
3160                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3161                 $text = $frame->expand( $dom, $flags );
3162
3163                 wfProfileOut( $fname );
3164                 return $text;
3165         }
3166
3167         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3168         static function createAssocArgs( $args ) {
3169                 $assocArgs = array();
3170                 $index = 1;
3171                 foreach( $args as $arg ) {
3172                         $eqpos = strpos( $arg, '=' );
3173                         if ( $eqpos === false ) {
3174                                 $assocArgs[$index++] = $arg;
3175                         } else {
3176                                 $name = trim( substr( $arg, 0, $eqpos ) );
3177                                 $value = trim( substr( $arg, $eqpos+1 ) );
3178                                 if ( $value === false ) {
3179                                         $value = '';
3180                                 }
3181                                 if ( $name !== false ) {
3182                                         $assocArgs[$name] = $value;
3183                                 }
3184                         }
3185                 }
3186
3187                 return $assocArgs;
3188         }
3189
3190         /**
3191          * Return the text of a template, after recursively
3192          * replacing any variables or templates within the template.
3193          *
3194          * @param array $piece The parts of the template
3195          *  $piece['text']: matched text
3196          *  $piece['title']: the title, i.e. the part before the |
3197          *  $piece['parts']: the parameter array
3198          * @param PPFrame The current frame, contains template arguments
3199          * @return string the text of the template
3200          * @private
3201          */
3202         function braceSubstitution( $piece, $frame ) {
3203                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3204                 $fname = __METHOD__;
3205                 wfProfileIn( $fname );
3206                 wfProfileIn( __METHOD__.'-setup' );
3207
3208                 # Flags
3209                 $found = false;             # $text has been filled
3210                 $nowiki = false;            # wiki markup in $text should be escaped
3211                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3212                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3213                 $isDOM = false;             # $text is a DOM node needing expansion
3214
3215                 # Title object, where $text came from
3216                 $title = NULL;
3217
3218                 # $part1 is the bit before the first |, and must contain only title characters.
3219                 # Various prefixes will be stripped from it later.
3220                 $titleWithSpaces = $frame->expand( $piece['title'] );
3221                 $part1 = trim( $titleWithSpaces );
3222                 $titleText = false;
3223
3224                 # Original title text preserved for various purposes
3225                 $originalTitle = $part1;
3226
3227                 # $args is a list of argument nodes, starting from index 0, not including $part1
3228                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3229                 wfProfileOut( __METHOD__.'-setup' );
3230
3231                 # SUBST
3232                 wfProfileIn( __METHOD__.'-modifiers' );
3233                 if ( !$found ) {
3234                         $mwSubst =& MagicWord::get( 'subst' );
3235                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3236                                 # One of two possibilities is true:
3237                                 # 1) Found SUBST but not in the PST phase
3238                                 # 2) Didn't find SUBST and in the PST phase
3239                                 # In either case, return without further processing
3240                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3241                                 $found = true;
3242                         }
3243                 }
3244
3245                 # Variables
3246                 if ( !$found && $args->length == 0 ) {
3247                         $id = $this->mVariables->matchStartToEnd( $part1 );
3248                         if ( $id !== false ) {
3249                                 $text = $this->getVariableValue( $id );
3250                                 if (MagicWord::getCacheTTL($id)>-1)
3251                                         $this->mOutput->mContainsOldMagic = true;
3252                                 $found = true;
3253                         }
3254                 }
3255
3256                 # MSG, MSGNW and RAW
3257                 if ( !$found ) {
3258                         # Check for MSGNW:
3259                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3260                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3261                                 $nowiki = true;
3262                         } else {
3263                                 # Remove obsolete MSG:
3264                                 $mwMsg =& MagicWord::get( 'msg' );
3265                                 $mwMsg->matchStartAndRemove( $part1 );
3266                         }
3267
3268                         # Check for RAW:
3269                         $mwRaw =& MagicWord::get( 'raw' );
3270                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3271                                 $forceRawInterwiki = true;
3272                         }
3273                 }
3274                 wfProfileOut( __METHOD__.'-modifiers' );
3275
3276                 # Parser functions
3277                 if ( !$found ) {
3278                         wfProfileIn( __METHOD__ . '-pfunc' );
3279
3280                         $colonPos = strpos( $part1, ':' );
3281                         if ( $colonPos !== false ) {
3282                                 # Case sensitive functions
3283                                 $function = substr( $part1, 0, $colonPos );
3284                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3285                                         $function = $this->mFunctionSynonyms[1][$function];
3286                                 } else {
3287                                         # Case insensitive functions
3288                                         $function = strtolower( $function );
3289                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3290                                                 $function = $this->mFunctionSynonyms[0][$function];
3291                                         } else {
3292                                                 $function = false;
3293                                         }
3294                                 }
3295                                 if ( $function ) {
3296                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3297                                         $initialArgs = array( &$this );
3298                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3299                                         if ( $flags & SFH_OBJECT_ARGS ) {
3300                                                 # Add a frame parameter, and pass the arguments as an array
3301                                                 $allArgs = $initialArgs;
3302                                                 $allArgs[] = $frame;
3303                                                 foreach ( $args as $arg ) {
3304                                                         $funcArgs[] = $arg;
3305                                                 }
3306                                                 $allArgs[] = $funcArgs;
3307                                         } else {
3308                                                 # Convert arguments to plain text
3309                                                 foreach ( $args as $arg ) {
3310                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3311                                                 }
3312                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3313                                         }
3314
3315                                         $result = call_user_func_array( $callback, $allArgs );
3316                                         $found = true;
3317
3318                                         if ( is_array( $result ) ) {
3319                                                 if ( isset( $result[0] ) ) {
3320                                                         $text = $result[0];
3321                                                         unset( $result[0] );
3322                                                 }
3323
3324                                                 // Extract flags into the local scope
3325                                                 // This allows callers to set flags such as nowiki, found, etc.
3326                                                 extract( $result );
3327                                         } else {
3328                                                 $text = $result;
3329                                         }
3330                                 }
3331                         }
3332                         wfProfileOut( __METHOD__ . '-pfunc' );
3333                 }
3334
3335                 # Finish mangling title and then check for loops.
3336                 # Set $title to a Title object and $titleText to the PDBK
3337                 if ( !$found ) {
3338                         $ns = NS_TEMPLATE;
3339                         # Split the title into page and subpage
3340                         $subpage = '';
3341                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3342                         if ($subpage !== '') {
3343                                 $ns = $this->mTitle->getNamespace();
3344                         }
3345                         $title = Title::newFromText( $part1, $ns );
3346                         if ( $title ) {
3347                                 $titleText = $title->getPrefixedText();
3348                                 # Check for language variants if the template is not found
3349                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3350                                         $wgContLang->findVariantLink($part1, $title);
3351                                 }
3352                                 # Do infinite loop check
3353                                 if ( isset( $frame->loopCheckHash[$titleText] ) ) {
3354                                         $found = true;
3355                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
3356                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3357                                 }
3358                                 # Do recursion depth check
3359                                 $limit = $this->mOptions->getMaxTemplateDepth();
3360                                 if ( $frame->depth >= $limit ) {
3361                                         $found = true;
3362                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
3363                                 }
3364                         }
3365                 }
3366
3367                 # Load from database
3368                 if ( !$found && $title ) {
3369                         wfProfileIn( __METHOD__ . '-loadtpl' );
3370                         if ( !$title->isExternal() ) {
3371                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3372                                         $text = SpecialPage::capturePath( $title );
3373                                         if ( is_string( $text ) ) {
3374                                                 $found = true;
3375                                                 $isHTML = true;
3376                                                 $this->disableCache();
3377                                         }
3378                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3379                                         $found = false; //access denied
3380                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3381                                 } else {
3382                                         list( $text, $title ) = $this->getTemplateDom( $title );
3383                                         if ( $text !== false ) {
3384                                                 $found = true;
3385                                                 $isDOM = true;
3386                                         }
3387                                 }
3388
3389                                 # If the title is valid but undisplayable, make a link to it
3390                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3391                                         $text = "[[:$titleText]]";
3392                                         $found = true;
3393                                 }
3394                         } elseif ( $title->isTrans() ) {
3395                                 // Interwiki transclusion
3396                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3397                                         $text = $this->interwikiTransclude( $title, 'render' );
3398                                         $isHTML = true;
3399                                 } else {
3400                                         $text = $this->interwikiTransclude( $title, 'raw' );
3401                                         // Preprocess it like a template
3402                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3403                                         $isDOM = true;
3404                                 }
3405                                 $found = true;
3406                         }
3407                         wfProfileOut( __METHOD__ . '-loadtpl' );
3408                 }
3409
3410                 # If we haven't found text to substitute by now, we're done
3411                 # Recover the source wikitext and return it
3412                 if ( !$found ) {
3413                         $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3414                         wfProfileOut( $fname );
3415                         return $text;
3416                 }
3417
3418                 # Expand DOM-style return values in a child frame
3419                 if ( $isDOM ) {
3420                         # Clean up argument array
3421                         $newFrame = $frame->newChild( $args, $title );
3422                         # Add a new element to the templace loop detection hashtable
3423                         $newFrame->loopCheckHash[$titleText] = true;
3424
3425                         if ( $titleText !== false && $newFrame->isEmpty() ) {
3426                                 # Expansion is eligible for the empty-frame cache
3427                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3428                                         $text = $this->mTplExpandCache[$titleText];
3429                                 } else {
3430                                         $text = $newFrame->expand( $text );
3431                                         $this->mTplExpandCache[$titleText] = $text;
3432                                 }
3433                         } else {
3434                                 # Uncached expansion
3435                                 $text = $newFrame->expand( $text );
3436                         }
3437                 }
3438
3439                 # Replace raw HTML by a placeholder
3440                 # Add a blank line preceding, to prevent it from mucking up
3441                 # immediately preceding headings
3442                 if ( $isHTML ) {
3443                         $text = "\n\n" . $this->insertStripItem( $text );
3444                 }
3445                 # Escape nowiki-style return values
3446                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3447                         $text = wfEscapeWikiText( $text );
3448                 }
3449                 # Bug 529: if the template begins with a table or block-level
3450                 # element, it should be treated as beginning a new line.
3451                 # This behaviour is somewhat controversial.
3452                 elseif ( !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3453                         $text = "\n" . $text;
3454                 }
3455
3456                 if ( !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3457                         # Error, oversize inclusion
3458                         $text = "[[$originalTitle]]" .
3459                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3460                 }
3461
3462                 wfProfileOut( $fname );
3463                 return $text;
3464         }
3465
3466         /**
3467          * Get the semi-parsed DOM representation of a template with a given title,
3468          * and its redirect destination title. Cached.
3469          */
3470         function getTemplateDom( $title ) {
3471                 $cacheTitle = $title;
3472                 $titleText = $title->getPrefixedDBkey();
3473
3474                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3475                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3476                         $title = Title::makeTitle( $ns, $dbk );
3477                         $titleText = $title->getPrefixedDBkey();
3478                 }
3479                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3480                         return array( $this->mTplDomCache[$titleText], $title );
3481                 }
3482
3483                 // Cache miss, go to the database
3484                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3485
3486                 if ( $text === false ) {
3487                         $this->mTplDomCache[$titleText] = false;
3488                         return array( false, $title );
3489                 }
3490
3491                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3492                 $this->mTplDomCache[ $titleText ] = $dom;
3493
3494                 if (! $title->equals($cacheTitle)) {
3495                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3496                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3497                 }
3498
3499                 return array( $dom, $title );
3500         }
3501
3502         /**
3503          * Fetch the unparsed text of a template and register a reference to it.
3504          */
3505         function fetchTemplateAndTitle( $title ) {
3506                 $templateCb = $this->mOptions->getTemplateCallback();
3507                 $stuff = call_user_func( $templateCb, $title );
3508                 $text = $stuff['text'];
3509                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3510                 if ( isset( $stuff['deps'] ) ) {
3511                         foreach ( $stuff['deps'] as $dep ) {
3512                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3513                         }
3514                 }
3515                 return array($text,$finalTitle);
3516         }
3517
3518         function fetchTemplate( $title ) {
3519                 $rv = $this->fetchTemplateAndTitle($title);
3520                 return $rv[0];
3521         }
3522
3523         /**
3524          * Static function to get a template
3525          * Can be overridden via ParserOptions::setTemplateCallback().
3526          */
3527         static function statelessFetchTemplate( $title ) {
3528                 $text = $skip = false;
3529                 $finalTitle = $title;
3530                 $deps = array();
3531
3532                 // Loop to fetch the article, with up to 1 redirect
3533                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3534                         # Give extensions a chance to select the revision instead
3535                         $id = false; // Assume current
3536                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3537
3538                         if( $skip ) {
3539                                 $text = false;
3540                                 $deps[] = array(
3541                                         'title' => $title,
3542                                         'page_id' => $title->getArticleID(),
3543                                         'rev_id' => null );
3544                                 break;
3545                         }
3546                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3547                         $rev_id = $rev ? $rev->getId() : 0;
3548
3549                         $deps[] = array(
3550                                 'title' => $title,
3551                                 'page_id' => $title->getArticleID(),
3552                                 'rev_id' => $rev_id );
3553
3554                         if( $rev ) {
3555                                 $text = $rev->getText();
3556                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3557                                 global $wgLang;
3558                                 $message = $wgLang->lcfirst( $title->getText() );
3559                                 $text = wfMsgForContentNoTrans( $message );
3560                                 if( wfEmptyMsg( $message, $text ) ) {
3561                                         $text = false;
3562                                         break;
3563                                 }
3564                         } else {
3565                                 break;
3566                         }
3567                         if ( $text === false ) {
3568                                 break;
3569                         }
3570                         // Redirect?
3571                         $finalTitle = $title;
3572                         $title = Title::newFromRedirect( $text );
3573                 }
3574                 return array(
3575                         'text' => $text,
3576                         'finalTitle' => $finalTitle,
3577                         'deps' => $deps );
3578         }
3579
3580         /**
3581          * Transclude an interwiki link.
3582          */
3583         function interwikiTransclude( $title, $action ) {
3584                 global $wgEnableScaryTranscluding;
3585
3586                 if (!$wgEnableScaryTranscluding)
3587                         return wfMsg('scarytranscludedisabled');
3588
3589                 $url = $title->getFullUrl( "action=$action" );
3590
3591                 if (strlen($url) > 255)
3592                         return wfMsg('scarytranscludetoolong');
3593                 return $this->fetchScaryTemplateMaybeFromCache($url);
3594         }
3595
3596         function fetchScaryTemplateMaybeFromCache($url) {
3597                 global $wgTranscludeCacheExpiry;
3598                 $dbr = wfGetDB(DB_SLAVE);
3599                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3600                                 array('tc_url' => $url));
3601                 if ($obj) {
3602                         $time = $obj->tc_time;
3603                         $text = $obj->tc_contents;
3604                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3605                                 return $text;
3606                         }
3607                 }
3608
3609                 $text = Http::get($url);
3610                 if (!$text)
3611                         return wfMsg('scarytranscludefailed', $url);
3612
3613                 $dbw = wfGetDB(DB_MASTER);
3614                 $dbw->replace('transcache', array('tc_url'), array(
3615                         'tc_url' => $url,
3616                         'tc_time' => time(),
3617                         'tc_contents' => $text));
3618                 return $text;
3619         }
3620
3621
3622         /**
3623          * Triple brace replacement -- used for template arguments
3624          * @private
3625          */
3626         function argSubstitution( $piece, $frame ) {
3627                 wfProfileIn( __METHOD__ );
3628
3629                 $error = false;
3630                 $parts = $piece['parts'];
3631                 $nameWithSpaces = $frame->expand( $piece['title'] );
3632                 $argName = trim( $nameWithSpaces );
3633
3634                 $text = $frame->getArgument( $argName );
3635                 if (  $text === false && ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3636                         # No match in frame, use the supplied default
3637                         $text = $frame->expand( $parts->item( 0 ) );
3638                 }
3639                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3640                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3641                 }
3642
3643                 if ( $text === false ) {
3644                         # No match anywhere
3645                         $text = '{{{' . $frame->implode( '|', $nameWithSpaces, $parts ) . '}}}';
3646                 }
3647                 if ( $error !== false ) {
3648                         $text .= $error;
3649                 }
3650
3651                 wfProfileOut( __METHOD__ );
3652                 return $text;
3653         }
3654
3655         /**
3656          * Return the text to be used for a given extension tag.
3657          * This is the ghost of strip().
3658          *
3659          * @param array $params Associative array of parameters:
3660          *     name       DOMNode for the tag name
3661          *     attr       DOMNode for unparsed text where tag attributes are thought to be
3662          *     attributes Optional associative array of parsed attributes
3663          *     inner      Contents of extension element
3664          *     noClose    Original text did not have a close tag
3665          * @param PPFrame $frame
3666          */
3667         function extensionSubstitution( $params, $frame ) {
3668                 global $wgRawHtml, $wgContLang;
3669                 static $n = 1;
3670
3671                 $name = $frame->expand( $params['name'] );
3672                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3673                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3674
3675                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3676
3677                 if ( $this->ot['html'] ) {
3678                         $name = strtolower( $name );
3679
3680                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3681                         if ( isset( $params['attributes'] ) ) {
3682                                 $attributes = $attributes + $params['attributes'];
3683                         }
3684                         switch ( $name ) {
3685                                 case 'html':
3686                                         if( $wgRawHtml ) {
3687                                                 $output = $content;
3688                                                 break;
3689                                         } else {
3690                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3691                                         }
3692                                 case 'nowiki':
3693                                         $output = Xml::escapeTagsOnly( $content );
3694                                         break;
3695                                 case 'math':
3696                                         $output = $wgContLang->armourMath(
3697                                                 MathRenderer::renderMath( $content, $attributes ) );
3698                                         break;
3699                                 case 'gallery':
3700                                         $output = $this->renderImageGallery( $content, $attributes );
3701                                         break;
3702                                 default:
3703                                         if( isset( $this->mTagHooks[$name] ) ) {
3704                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3705                                                         array( $content, $attributes, $this ) );
3706                                         } else {
3707                                                 throw new MWException( "Invalid call hook $name" );
3708                                         }
3709                         }
3710                 } else {
3711                         if ( $content === null ) {
3712                                 $output = "<$name$attrText/>";
3713                         } else {
3714                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3715                                 $output = "<$name$attrText>$content$close";
3716                         }
3717                 }
3718
3719                 if ( $name == 'html' || $name == 'nowiki' ) {
3720                         $this->mStripState->nowiki->setPair( $marker, $output );
3721                 } else {
3722                         $this->mStripState->general->setPair( $marker, $output );
3723                 }
3724                 return $marker;
3725         }
3726
3727         /**
3728          * Increment an include size counter
3729          *
3730          * @param string $type The type of expansion
3731          * @param integer $size The size of the text
3732          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3733          */
3734         function incrementIncludeSize( $type, $size ) {
3735                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3736                         return false;
3737                 } else {
3738                         $this->mIncludeSizes[$type] += $size;
3739                         return true;
3740                 }
3741         }
3742
3743         /**
3744          * Detect __NOGALLERY__ magic word and set a placeholder
3745          */
3746         function stripNoGallery( &$text ) {
3747                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3748                 # do not add TOC
3749                 $mw = MagicWord::get( 'nogallery' );
3750                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3751         }
3752
3753         /**
3754          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3755          * placeholder that will then be replaced by the real TOC in
3756          * ->formatHeadings, this works because at this points real
3757          * comments will have already been discarded by the sanitizer.
3758          *
3759          * Any additional __TOC__ magic words left over will be discarded
3760          * as there can only be one TOC on the page.
3761          */
3762         function stripToc( $text ) {
3763                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3764                 # do not add TOC
3765                 $mw = MagicWord::get( 'notoc' );
3766                 if( $mw->matchAndRemove( $text ) ) {
3767                         $this->mShowToc = false;
3768                 }
3769
3770                 $mw = MagicWord::get( 'toc' );
3771                 if( $mw->match( $text ) ) {
3772                         $this->mShowToc = true;
3773                         $this->mForceTocPosition = true;
3774
3775                         // Set a placeholder. At the end we'll fill it in with the TOC.
3776                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3777
3778                         // Only keep the first one.
3779                         $text = $mw->replace( '', $text );
3780                 }
3781                 return $text;
3782         }
3783
3784         /**
3785          * This function accomplishes several tasks:
3786          * 1) Auto-number headings if that option is enabled
3787          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3788          * 3) Add a Table of contents on the top for users who have enabled the option
3789          * 4) Auto-anchor headings
3790          *
3791          * It loops through all headlines, collects the necessary data, then splits up the
3792          * string and re-inserts the newly formatted headlines.
3793          *
3794          * @param string $text
3795          * @param boolean $isMain
3796          * @private
3797          */
3798         function formatHeadings( $text, $isMain=true ) {
3799                 global $wgMaxTocLevel, $wgContLang;
3800
3801                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3802                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3803                         $showEditLink = 0;
3804                 } else {
3805                         $showEditLink = $this->mOptions->getEditSection();
3806                 }
3807
3808                 # Inhibit editsection links if requested in the page
3809                 $esw =& MagicWord::get( 'noeditsection' );
3810                 if( $esw->matchAndRemove( $text ) ) {
3811                         $showEditLink = 0;
3812                 }
3813
3814                 # Get all headlines for numbering them and adding funky stuff like [edit]
3815                 # links - this is for later, but we need the number of headlines right now
3816                 $matches = array();
3817                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3818
3819                 # if there are fewer than 4 headlines in the article, do not show TOC
3820                 # unless it's been explicitly enabled.
3821                 $enoughToc = $this->mShowToc &&
3822                         (($numMatches >= 4) || $this->mForceTocPosition);
3823
3824                 # Allow user to stipulate that a page should have a "new section"
3825                 # link added via __NEWSECTIONLINK__
3826                 $mw =& MagicWord::get( 'newsectionlink' );
3827                 if( $mw->matchAndRemove( $text ) )
3828                         $this->mOutput->setNewSection( true );
3829
3830                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3831                 # override above conditions and always show TOC above first header
3832                 $mw =& MagicWord::get( 'forcetoc' );
3833                 if ($mw->matchAndRemove( $text ) ) {
3834                         $this->mShowToc = true;
3835                         $enoughToc = true;
3836                 }
3837
3838                 # We need this to perform operations on the HTML
3839                 $sk = $this->mOptions->getSkin();
3840
3841                 # headline counter
3842                 $headlineCount = 0;
3843                 $numVisible = 0;
3844
3845                 # Ugh .. the TOC should have neat indentation levels which can be
3846                 # passed to the skin functions. These are determined here
3847                 $toc = '';
3848                 $full = '';
3849                 $head = array();
3850                 $sublevelCount = array();
3851                 $levelCount = array();
3852                 $toclevel = 0;
3853                 $level = 0;
3854                 $prevlevel = 0;
3855                 $toclevel = 0;
3856                 $prevtoclevel = 0;
3857                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3858                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3859                 $tocraw = array();
3860
3861                 foreach( $matches[3] as $headline ) {
3862                         $isTemplate = false;
3863                         $titleText = false;
3864                         $sectionIndex = false;
3865                         $numbering = '';
3866                         $markerMatches = array();
3867                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3868                                 $serial = $markerMatches[1];
3869                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3870                                 $isTemplate = ($titleText != $baseTitleText);
3871                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3872                         }
3873
3874                         if( $toclevel ) {
3875                                 $prevlevel = $level;
3876                                 $prevtoclevel = $toclevel;
3877                         }
3878                         $level = $matches[1][$headlineCount];
3879
3880                         if( $doNumberHeadings || $enoughToc ) {
3881
3882                                 if ( $level > $prevlevel ) {
3883                                         # Increase TOC level
3884                                         $toclevel++;
3885                                         $sublevelCount[$toclevel] = 0;
3886                                         if( $toclevel<$wgMaxTocLevel ) {
3887                                                 $prevtoclevel = $toclevel;
3888                                                 $toc .= $sk->tocIndent();
3889                                                 $numVisible++;
3890                                         }
3891                                 }
3892                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3893                                         # Decrease TOC level, find level to jump to
3894
3895                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3896                                                 # Can only go down to level 1
3897                                                 $toclevel = 1;
3898                                         } else {
3899                                                 for ($i = $toclevel; $i > 0; $i--) {
3900                                                         if ( $levelCount[$i] == $level ) {
3901                                                                 # Found last matching level
3902                                                                 $toclevel = $i;
3903                                                                 break;
3904                                                         }
3905                                                         elseif ( $levelCount[$i] < $level ) {
3906                                                                 # Found first matching level below current level
3907                                                                 $toclevel = $i + 1;
3908                                                                 break;
3909                                                         }
3910                                                 }
3911                                         }
3912                                         if( $toclevel<$wgMaxTocLevel ) {
3913                                                 if($prevtoclevel < $wgMaxTocLevel) {
3914                                                         # Unindent only if the previous toc level was shown :p
3915                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3916                                                 } else {
3917                                                         $toc .= $sk->tocLineEnd();
3918                                                 }
3919                                         }
3920                                 }
3921                                 else {
3922                                         # No change in level, end TOC line
3923                                         if( $toclevel<$wgMaxTocLevel ) {
3924                                                 $toc .= $sk->tocLineEnd();
3925                                         }
3926                                 }
3927
3928                                 $levelCount[$toclevel] = $level;
3929
3930                                 # count number of headlines for each level
3931                                 @$sublevelCount[$toclevel]++;
3932                                 $dot = 0;
3933                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3934                                         if( !empty( $sublevelCount[$i] ) ) {
3935                                                 if( $dot ) {
3936                                                         $numbering .= '.';
3937                                                 }
3938                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3939                                                 $dot = 1;
3940                                         }
3941                                 }
3942                         }
3943
3944                         # The safe header is a version of the header text safe to use for links
3945                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3946                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3947
3948                         # Remove link placeholders by the link text.
3949                         #     <!--LINK number-->
3950                         # turns into
3951                         #     link text with suffix
3952                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3953                                                             "\$this->mLinkHolders['texts'][\$1]",
3954                                                             $safeHeadline );
3955                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3956                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3957                                                             $safeHeadline );
3958
3959                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3960                         $tocline = preg_replace(
3961                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3962                                 array( '',                          '<$1>'),
3963                                 $safeHeadline
3964                         );
3965                         $tocline = trim( $tocline );
3966
3967                         # For the anchor, strip out HTML-y stuff period
3968                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3969                         $safeHeadline = trim( $safeHeadline );
3970
3971                         # Save headline for section edit hint before it's escaped
3972                         $headlineHint = $safeHeadline;
3973                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3974                         $refers[$headlineCount] = $safeHeadline;
3975
3976                         # count how many in assoc. array so we can track dupes in anchors
3977                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3978                         $refcount[$headlineCount] = $refers[$safeHeadline];
3979
3980                         # Don't number the heading if it is the only one (looks silly)
3981                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3982                                 # the two are different if the line contains a link
3983                                 $headline=$numbering . ' ' . $headline;
3984                         }
3985
3986                         # Create the anchor for linking from the TOC to the section
3987                         $anchor = $safeHeadline;
3988                         if($refcount[$headlineCount] > 1 ) {
3989                                 $anchor .= '_' . $refcount[$headlineCount];
3990                         }
3991                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3992                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3993                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3994                         }
3995                         # give headline the correct <h#> tag
3996                         if( $showEditLink && $sectionIndex !== false ) {
3997                                 if( $isTemplate ) {
3998                                         # Put a T flag in the section identifier, to indicate to extractSections()
3999                                         # that sections inside <includeonly> should be counted.
4000                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
4001                                 } else {
4002                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
4003                                 }
4004                         } else {
4005                                 $editlink = '';
4006                         }
4007                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
4008
4009                         $headlineCount++;
4010                 }
4011
4012                 $this->mOutput->setSections( $tocraw );
4013
4014                 # Never ever show TOC if no headers
4015                 if( $numVisible < 1 ) {
4016                         $enoughToc = false;
4017                 }
4018
4019                 if( $enoughToc ) {
4020                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4021                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
4022                         }
4023                         $toc = $sk->tocList( $toc );
4024                 }
4025
4026                 # split up and insert constructed headlines
4027
4028                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4029                 $i = 0;
4030
4031                 foreach( $blocks as $block ) {
4032                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
4033                                 # This is the [edit] link that appears for the top block of text when
4034                                 # section editing is enabled
4035
4036                                 # Disabled because it broke block formatting
4037                                 # For example, a bullet point in the top line
4038                                 # $full .= $sk->editSectionLink(0);
4039                         }
4040                         $full .= $block;
4041                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4042                                 # Top anchor now in skin
4043                                 $full = $full.$toc;
4044                         }
4045
4046                         if( !empty( $head[$i] ) ) {
4047                                 $full .= $head[$i];
4048                         }
4049                         $i++;
4050                 }
4051                 if( $this->mForceTocPosition ) {
4052                         return str_replace( '<!--MWTOC-->', $toc, $full );
4053                 } else {
4054                         return $full;
4055                 }
4056         }
4057
4058         /**
4059          * Transform wiki markup when saving a page by doing \r\n -> \n
4060          * conversion, substitting signatures, {{subst:}} templates, etc.
4061          *
4062          * @param string $text the text to transform
4063          * @param Title &$title the Title object for the current article
4064          * @param User &$user the User object describing the current user
4065          * @param ParserOptions $options parsing options
4066          * @param bool $clearState whether to clear the parser state first
4067          * @return string the altered wiki markup
4068          * @public
4069          */
4070         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
4071                 $this->mOptions = $options;
4072                 $this->mTitle =& $title;
4073                 $this->setOutputType( OT_WIKI );
4074
4075                 if ( $clearState ) {
4076                         $this->clearState();
4077                 }
4078
4079                 $pairs = array(
4080                         "\r\n" => "\n",
4081                 );
4082                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4083                 $text = $this->pstPass2( $text, $user );
4084                 $text = $this->mStripState->unstripBoth( $text );
4085                 return $text;
4086         }
4087
4088         /**
4089          * Pre-save transform helper function
4090          * @private
4091          */
4092         function pstPass2( $text, $user ) {
4093                 global $wgContLang, $wgLocaltimezone;
4094
4095                 /* Note: This is the timestamp saved as hardcoded wikitext to
4096                  * the database, we use $wgContLang here in order to give
4097                  * everyone the same signature and use the default one rather
4098                  * than the one selected in each user's preferences.
4099                  */
4100                 if ( isset( $wgLocaltimezone ) ) {
4101                         $oldtz = getenv( 'TZ' );
4102                         putenv( 'TZ='.$wgLocaltimezone );
4103                 }
4104                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4105                   ' (' . date( 'T' ) . ')';
4106                 if ( isset( $wgLocaltimezone ) ) {
4107                         putenv( 'TZ='.$oldtz );
4108                 }
4109
4110                 # Variable replacement
4111                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4112                 $text = $this->replaceVariables( $text );
4113
4114                 # Strip out <nowiki> etc. added via replaceVariables
4115                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4116
4117                 # Signatures
4118                 $sigText = $this->getUserSig( $user );
4119                 $text = strtr( $text, array(
4120                         '~~~~~' => $d,
4121                         '~~~~' => "$sigText $d",
4122                         '~~~' => $sigText
4123                 ) );
4124
4125                 # Context links: [[|name]] and [[name (context)|]]
4126                 #
4127                 global $wgLegalTitleChars;
4128                 $tc = "[$wgLegalTitleChars]";
4129                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4130
4131                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4132                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4133                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4134
4135                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4136                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4137                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4138
4139                 $t = $this->mTitle->getText();
4140                 $m = array();
4141                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4142                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4143                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4144                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4145                 } else {
4146                         # if there's no context, don't bother duplicating the title
4147                         $text = preg_replace( $p2, '[[\\1]]', $text );
4148                 }
4149
4150                 # Trim trailing whitespace
4151                 $text = rtrim( $text );
4152
4153                 return $text;
4154         }
4155
4156         /**
4157          * Fetch the user's signature text, if any, and normalize to
4158          * validated, ready-to-insert wikitext.
4159          *
4160          * @param User $user
4161          * @return string
4162          * @private
4163          */
4164         function getUserSig( &$user ) {
4165                 global $wgMaxSigChars;
4166
4167                 $username = $user->getName();
4168                 $nickname = $user->getOption( 'nickname' );
4169                 $nickname = $nickname === '' ? $username : $nickname;
4170
4171                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4172                         $nickname = $username;
4173                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4174                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4175                         # Sig. might contain markup; validate this
4176                         if( $this->validateSig( $nickname ) !== false ) {
4177                                 # Validated; clean up (if needed) and return it
4178                                 return $this->cleanSig( $nickname, true );
4179                         } else {
4180                                 # Failed to validate; fall back to the default
4181                                 $nickname = $username;
4182                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4183                         }
4184                 }
4185
4186                 // Make sure nickname doesnt get a sig in a sig
4187                 $nickname = $this->cleanSigInSig( $nickname );
4188
4189                 # If we're still here, make it a link to the user page
4190                 $userText = wfEscapeWikiText( $username );
4191                 $nickText = wfEscapeWikiText( $nickname );
4192                 if ( $user->isAnon() )  {
4193                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4194                 } else {
4195                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4196                 }
4197         }
4198
4199         /**
4200          * Check that the user's signature contains no bad XML
4201          *
4202          * @param string $text
4203          * @return mixed An expanded string, or false if invalid.
4204          */
4205         function validateSig( $text ) {
4206                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4207         }
4208
4209         /**
4210          * Clean up signature text
4211          *
4212          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4213          * 2) Substitute all transclusions
4214          *
4215          * @param string $text
4216          * @param $parsing Whether we're cleaning (preferences save) or parsing
4217          * @return string Signature text
4218          */
4219         function cleanSig( $text, $parsing = false ) {
4220                 if ( !$parsing ) {
4221                         global $wgTitle;
4222                         $this->startExternalParse( $wgTitle, new ParserOptions(), OT_MSG );
4223                 }
4224
4225                 # FIXME: regex doesn't respect extension tags or nowiki
4226                 #  => Move this logic to braceSubstitution()
4227                 $substWord = MagicWord::get( 'subst' );
4228                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4229                 $substText = '{{' . $substWord->getSynonym( 0 );
4230
4231                 $text = preg_replace( $substRegex, $substText, $text );
4232                 $text = $this->cleanSigInSig( $text );
4233                 $dom = $this->preprocessToDom( $text );
4234                 $frame = new PPFrame( $this );
4235                 $text = $frame->expand( $dom->documentElement );
4236
4237                 if ( !$parsing ) {
4238                         $text = $this->mStripState->unstripBoth( $text );
4239                 }
4240
4241                 return $text;
4242         }
4243
4244         /**
4245          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4246          * @param string $text
4247          * @return string Signature text with /~{3,5}/ removed
4248          */
4249         function cleanSigInSig( $text ) {
4250                 $text = preg_replace( '/~{3,5}/', '', $text );
4251                 return $text;
4252         }
4253
4254         /**
4255          * Set up some variables which are usually set up in parse()
4256          * so that an external function can call some class members with confidence
4257          * @public
4258          */
4259         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4260                 $this->mTitle =& $title;
4261                 $this->mOptions = $options;
4262                 $this->setOutputType( $outputType );
4263                 if ( $clearState ) {
4264                         $this->clearState();
4265                 }
4266         }
4267
4268         /**
4269          * Transform a MediaWiki message by replacing magic variables.
4270          *
4271          * For some unknown reason, it also expands templates, but only to the
4272          * first recursion level. This is wrong and broken, probably introduced
4273          * accidentally during refactoring, but probably relied upon by thousands
4274          * of users.
4275          *
4276          * @param string $text the text to transform
4277          * @param ParserOptions $options  options
4278          * @return string the text with variables substituted
4279          * @public
4280          */
4281         function transformMsg( $text, $options ) {
4282                 global $wgTitle;
4283                 static $executing = false;
4284
4285                 $fname = "Parser::transformMsg";
4286
4287                 # Guard against infinite recursion
4288                 if ( $executing ) {
4289                         return $text;
4290                 }
4291                 $executing = true;
4292
4293                 wfProfileIn($fname);
4294
4295                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4296                         $this->mTitle = $wgTitle;
4297                 } else {
4298                         $this->mTitle = Title::newFromText('msg');
4299                 }
4300                 $this->mOptions = $options;
4301                 $this->setOutputType( OT_MSG );
4302                 $this->clearState();
4303                 $text = $this->replaceVariables( $text );
4304                 $text = $this->mStripState->unstripBoth( $text );
4305
4306                 $executing = false;
4307                 wfProfileOut($fname);
4308                 return $text;
4309         }
4310
4311         /**
4312          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4313          * The callback should have the following form:
4314          *    function myParserHook( $text, $params, &$parser ) { ... }
4315          *
4316          * Transform and return $text. Use $parser for any required context, e.g. use
4317          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4318          *
4319          * @public
4320          *
4321          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4322          * @param mixed $callback The callback function (and object) to use for the tag
4323          *
4324          * @return The old value of the mTagHooks array associated with the hook
4325          */
4326         function setHook( $tag, $callback ) {
4327                 $tag = strtolower( $tag );
4328                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4329                 $this->mTagHooks[$tag] = $callback;
4330                 $this->mStripList[] = $tag;
4331
4332                 return $oldVal;
4333         }
4334
4335         function setTransparentTagHook( $tag, $callback ) {
4336                 $tag = strtolower( $tag );
4337                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4338                 $this->mTransparentTagHooks[$tag] = $callback;
4339
4340                 return $oldVal;
4341         }
4342
4343         /**
4344          * Create a function, e.g. {{sum:1|2|3}}
4345          * The callback function should have the form:
4346          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4347          *
4348          * The callback may either return the text result of the function, or an array with the text
4349          * in element 0, and a number of flags in the other elements. The names of the flags are
4350          * specified in the keys. Valid flags are:
4351          *   found                     The text returned is valid, stop processing the template. This
4352          *                             is on by default.
4353          *   nowiki                    Wiki markup in the return value should be escaped
4354          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4355          *
4356          * @public
4357          *
4358          * @param string $id The magic word ID
4359          * @param mixed $callback The callback function (and object) to use
4360          * @param integer $flags a combination of the following flags:
4361          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4362          *
4363          * @return The old callback function for this name, if any
4364          */
4365         function setFunctionHook( $id, $callback, $flags = 0 ) {
4366                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4367                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4368
4369                 # Add to function cache
4370                 $mw = MagicWord::get( $id );
4371                 if( !$mw )
4372                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4373
4374                 $synonyms = $mw->getSynonyms();
4375                 $sensitive = intval( $mw->isCaseSensitive() );
4376
4377                 foreach ( $synonyms as $syn ) {
4378                         # Case
4379                         if ( !$sensitive ) {
4380                                 $syn = strtolower( $syn );
4381                         }
4382                         # Add leading hash
4383                         if ( !( $flags & SFH_NO_HASH ) ) {
4384                                 $syn = '#' . $syn;
4385                         }
4386                         # Remove trailing colon
4387                         if ( substr( $syn, -1, 1 ) == ':' ) {
4388                                 $syn = substr( $syn, 0, -1 );
4389                         }
4390                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4391                 }
4392                 return $oldVal;
4393         }
4394
4395         /**
4396          * Get all registered function hook identifiers
4397          *
4398          * @return array
4399          */
4400         function getFunctionHooks() {
4401                 return array_keys( $this->mFunctionHooks );
4402         }
4403
4404         /**
4405          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4406          * Placeholders created in Skin::makeLinkObj()
4407          * Returns an array of link CSS classes, indexed by PDBK.
4408          * $options is a bit field, RLH_FOR_UPDATE to select for update
4409          */
4410         function replaceLinkHolders( &$text, $options = 0 ) {
4411                 global $wgUser;
4412                 global $wgContLang;
4413
4414                 $fname = 'Parser::replaceLinkHolders';
4415                 wfProfileIn( $fname );
4416
4417                 $pdbks = array();
4418                 $colours = array();
4419                 $linkcolour_ids = array();
4420                 $sk = $this->mOptions->getSkin();
4421                 $linkCache =& LinkCache::singleton();
4422
4423                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4424                         wfProfileIn( $fname.'-check' );
4425                         $dbr = wfGetDB( DB_SLAVE );
4426                         $page = $dbr->tableName( 'page' );
4427                         $threshold = $wgUser->getOption('stubthreshold');
4428
4429                         # Sort by namespace
4430                         asort( $this->mLinkHolders['namespaces'] );
4431
4432                         # Generate query
4433                         $query = false;
4434                         $current = null;
4435                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4436                                 # Make title object
4437                                 $title = $this->mLinkHolders['titles'][$key];
4438
4439                                 # Skip invalid entries.
4440                                 # Result will be ugly, but prevents crash.
4441                                 if ( is_null( $title ) ) {
4442                                         continue;
4443                                 }
4444                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4445
4446                                 # Check if it's a static known link, e.g. interwiki
4447                                 if ( $title->isAlwaysKnown() ) {
4448                                         $colours[$pdbk] = '';
4449                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4450                                         $colours[$pdbk] = '';
4451                                         $this->mOutput->addLink( $title, $id );
4452                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4453                                         $colours[$pdbk] = 'new';
4454                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4455                                         $colours[$pdbk] = 'new';
4456                                 } else {
4457                                         # Not in the link cache, add it to the query
4458                                         if ( !isset( $current ) ) {
4459                                                 $current = $ns;
4460                                                 $query =  "SELECT page_id, page_namespace, page_title";
4461                                                 if ( $threshold > 0 ) {
4462                                                         $query .= ', page_len, page_is_redirect';
4463                                                 }
4464                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4465                                         } elseif ( $current != $ns ) {
4466                                                 $current = $ns;
4467                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4468                                         } else {
4469                                                 $query .= ', ';
4470                                         }
4471
4472                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4473                                 }
4474                         }
4475                         if ( $query ) {
4476                                 $query .= '))';
4477                                 if ( $options & RLH_FOR_UPDATE ) {
4478                                         $query .= ' FOR UPDATE';
4479                                 }
4480
4481                                 $res = $dbr->query( $query, $fname );
4482
4483                                 # Fetch data and form into an associative array
4484                                 # non-existent = broken
4485                                 while ( $s = $dbr->fetchObject($res) ) {
4486                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4487                                         $pdbk = $title->getPrefixedDBkey();
4488                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4489                                         $this->mOutput->addLink( $title, $s->page_id );
4490                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4491                                         //add id to the extension todolist
4492                                         $linkcolour_ids[$s->page_id] = $pdbk;
4493                                 }
4494                                 //pass an array of page_ids to an extension
4495                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4496                         }
4497                         wfProfileOut( $fname.'-check' );
4498
4499                         # Do a second query for different language variants of links and categories
4500                         if($wgContLang->hasVariants()){
4501                                 $linkBatch = new LinkBatch();
4502                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4503                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4504                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4505
4506                                 $categories = $this->mOutput->getCategoryLinks();
4507
4508                                 // Add variants of links to link batch
4509                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4510                                         $title = $this->mLinkHolders['titles'][$key];
4511                                         if ( is_null( $title ) )
4512                                                 continue;
4513
4514                                         $pdbk = $title->getPrefixedDBkey();
4515                                         $titleText = $title->getText();
4516
4517                                         // generate all variants of the link title text
4518                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4519
4520                                         // if link was not found (in first query), add all variants to query
4521                                         if ( !isset($colours[$pdbk]) ){
4522                                                 foreach($allTextVariants as $textVariant){
4523                                                         if($textVariant != $titleText){
4524                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4525                                                                 if(is_null($variantTitle)) continue;
4526                                                                 $linkBatch->addObj( $variantTitle );
4527                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4528                                                         }
4529                                                 }
4530                                         }
4531                                 }
4532
4533                                 // process categories, check if a category exists in some variant
4534                                 foreach( $categories as $category ){
4535                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4536                                         foreach($variants as $variant){
4537                                                 if($variant != $category){
4538                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4539                                                         if(is_null($variantTitle)) continue;
4540                                                         $linkBatch->addObj( $variantTitle );
4541                                                         $categoryMap[$variant] = $category;
4542                                                 }
4543                                         }
4544                                 }
4545
4546
4547                                 if(!$linkBatch->isEmpty()){
4548                                         // construct query
4549                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4550
4551                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4552                                         if ( $threshold > 0 ) {
4553                                                 $variantQuery .= ', page_len, page_is_redirect';
4554                                         }
4555
4556                                         $variantQuery .= " FROM $page WHERE $titleClause";
4557                                         if ( $options & RLH_FOR_UPDATE ) {
4558                                                 $variantQuery .= ' FOR UPDATE';
4559                                         }
4560
4561                                         $varRes = $dbr->query( $variantQuery, $fname );
4562
4563                                         // for each found variants, figure out link holders and replace
4564                                         while ( $s = $dbr->fetchObject($varRes) ) {
4565
4566                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4567                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4568                                                 $vardbk = $variantTitle->getDBkey();
4569
4570                                                 $holderKeys = array();
4571                                                 if(isset($variantMap[$varPdbk])){
4572                                                         $holderKeys = $variantMap[$varPdbk];
4573                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4574                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4575                                                 }
4576
4577                                                 // loop over link holders
4578                                                 foreach($holderKeys as $key){
4579                                                         $title = $this->mLinkHolders['titles'][$key];
4580                                                         if ( is_null( $title ) ) continue;
4581
4582                                                         $pdbk = $title->getPrefixedDBkey();
4583
4584                                                         if(!isset($colours[$pdbk])){
4585                                                                 // found link in some of the variants, replace the link holder data
4586                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4587                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4588
4589                                                                 // set pdbk and colour
4590                                                                 $pdbks[$key] = $varPdbk;
4591                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4592                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4593                                                         }
4594                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4595                                                 }
4596
4597                                                 // check if the object is a variant of a category
4598                                                 if(isset($categoryMap[$vardbk])){
4599                                                         $oldkey = $categoryMap[$vardbk];
4600                                                         if($oldkey != $vardbk)
4601                                                                 $varCategories[$oldkey]=$vardbk;
4602                                                 }
4603                                         }
4604
4605                                         // rebuild the categories in original order (if there are replacements)
4606                                         if(count($varCategories)>0){
4607                                                 $newCats = array();
4608                                                 $originalCats = $this->mOutput->getCategories();
4609                                                 foreach($originalCats as $cat => $sortkey){
4610                                                         // make the replacement
4611                                                         if( array_key_exists($cat,$varCategories) )
4612                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4613                                                         else $newCats[$cat] = $sortkey;
4614                                                 }
4615                                                 $this->mOutput->setCategoryLinks($newCats);
4616                                         }
4617                                 }
4618                         }
4619
4620                         # Construct search and replace arrays
4621                         wfProfileIn( $fname.'-construct' );
4622                         $replacePairs = array();
4623                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4624                                 $pdbk = $pdbks[$key];
4625                                 $searchkey = "<!--LINK $key-->";
4626                                 $title = $this->mLinkHolders['titles'][$key];
4627                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4628                                         $linkCache->addBadLinkObj( $title );
4629                                         $colours[$pdbk] = 'new';
4630                                         $this->mOutput->addLink( $title, 0 );
4631                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4632                                                                         $this->mLinkHolders['texts'][$key],
4633                                                                         $this->mLinkHolders['queries'][$key] );
4634                                 } else {
4635                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4636                                                                         $this->mLinkHolders['texts'][$key],
4637                                                                         $this->mLinkHolders['queries'][$key] );
4638                                 }
4639                         }
4640                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4641                         wfProfileOut( $fname.'-construct' );
4642
4643                         # Do the thing
4644                         wfProfileIn( $fname.'-replace' );
4645                         $text = preg_replace_callback(
4646                                 '/(<!--LINK .*?-->)/',
4647                                 $replacer->cb(),
4648                                 $text);
4649
4650                         wfProfileOut( $fname.'-replace' );
4651                 }
4652
4653                 # Now process interwiki link holders
4654                 # This is quite a bit simpler than internal links
4655                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4656                         wfProfileIn( $fname.'-interwiki' );
4657                         # Make interwiki link HTML
4658                         $replacePairs = array();
4659                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4660                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4661                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4662                         }
4663                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4664
4665                         $text = preg_replace_callback(
4666                                 '/<!--IWLINK (.*?)-->/',
4667                                 $replacer->cb(),
4668                                 $text );
4669                         wfProfileOut( $fname.'-interwiki' );
4670                 }
4671
4672                 wfProfileOut( $fname );
4673                 return $colours;
4674         }
4675
4676         /**
4677          * Replace <!--LINK--> link placeholders with plain text of links
4678          * (not HTML-formatted).
4679          * @param string $text
4680          * @return string
4681          */
4682         function replaceLinkHoldersText( $text ) {
4683                 $fname = 'Parser::replaceLinkHoldersText';
4684                 wfProfileIn( $fname );
4685
4686                 $text = preg_replace_callback(
4687                         '/<!--(LINK|IWLINK) (.*?)-->/',
4688                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4689                         $text );
4690
4691                 wfProfileOut( $fname );
4692                 return $text;
4693         }
4694
4695         /**
4696          * @param array $matches
4697          * @return string
4698          * @private
4699          */
4700         function replaceLinkHoldersTextCallback( $matches ) {
4701                 $type = $matches[1];
4702                 $key  = $matches[2];
4703                 if( $type == 'LINK' ) {
4704                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4705                                 return $this->mLinkHolders['texts'][$key];
4706                         }
4707                 } elseif( $type == 'IWLINK' ) {
4708                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4709                                 return $this->mInterwikiLinkHolders['texts'][$key];
4710                         }
4711                 }
4712                 return $matches[0];
4713         }
4714
4715         /**
4716          * Tag hook handler for 'pre'.
4717          */
4718         function renderPreTag( $text, $attribs ) {
4719                 // Backwards-compatibility hack
4720                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4721
4722                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4723                 return wfOpenElement( 'pre', $attribs ) .
4724                         Xml::escapeTagsOnly( $content ) .
4725                         '</pre>';
4726         }
4727
4728         /**
4729          * Renders an image gallery from a text with one line per image.
4730          * text labels may be given by using |-style alternative text. E.g.
4731          *   Image:one.jpg|The number "1"
4732          *   Image:tree.jpg|A tree
4733          * given as text will return the HTML of a gallery with two images,
4734          * labeled 'The number "1"' and
4735          * 'A tree'.
4736          */
4737         function renderImageGallery( $text, $params ) {
4738                 $ig = new ImageGallery();
4739                 $ig->setContextTitle( $this->mTitle );
4740                 $ig->setShowBytes( false );
4741                 $ig->setShowFilename( false );
4742                 $ig->setParser( $this );
4743                 $ig->setHideBadImages();
4744                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4745                 $ig->useSkin( $this->mOptions->getSkin() );
4746                 $ig->mRevisionId = $this->mRevisionId;
4747
4748                 if( isset( $params['caption'] ) ) {
4749                         $caption = $params['caption'];
4750                         $caption = htmlspecialchars( $caption );
4751                         $caption = $this->replaceInternalLinks( $caption );
4752                         $ig->setCaptionHtml( $caption );
4753                 }
4754                 if( isset( $params['perrow'] ) ) {
4755                         $ig->setPerRow( $params['perrow'] );
4756                 }
4757                 if( isset( $params['widths'] ) ) {
4758                         $ig->setWidths( $params['widths'] );
4759                 }
4760                 if( isset( $params['heights'] ) ) {
4761                         $ig->setHeights( $params['heights'] );
4762                 }
4763
4764                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4765
4766                 $lines = explode( "\n", $text );
4767                 foreach ( $lines as $line ) {
4768                         # match lines like these:
4769                         # Image:someimage.jpg|This is some image
4770                         $matches = array();
4771                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4772                         # Skip empty lines
4773                         if ( count( $matches ) == 0 ) {
4774                                 continue;
4775                         }
4776                         $tp = Title::newFromText( $matches[1] );
4777                         $nt =& $tp;
4778                         if( is_null( $nt ) ) {
4779                                 # Bogus title. Ignore these so we don't bomb out later.
4780                                 continue;
4781                         }
4782                         if ( isset( $matches[3] ) ) {
4783                                 $label = $matches[3];
4784                         } else {
4785                                 $label = '';
4786                         }
4787
4788                         $pout = $this->parse( $label,
4789                                 $this->mTitle,
4790                                 $this->mOptions,
4791                                 false, // Strip whitespace...?
4792                                 false  // Don't clear state!
4793                         );
4794                         $html = $pout->getText();
4795
4796                         $ig->add( $nt, $html );
4797
4798                         # Only add real images (bug #5586)
4799                         if ( $nt->getNamespace() == NS_IMAGE ) {
4800                                 $this->mOutput->addImage( $nt->getDBkey() );
4801                         }
4802                 }
4803                 return $ig->toHTML();
4804         }
4805
4806         function getImageParams( $handler ) {
4807                 if ( $handler ) {
4808                         $handlerClass = get_class( $handler );
4809                 } else {
4810                         $handlerClass = '';
4811                 }
4812                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4813                         // Initialise static lists
4814                         static $internalParamNames = array(
4815                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4816                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4817                                         'bottom', 'text-bottom' ),
4818                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4819                                         'upright', 'border' ),
4820                         );
4821                         static $internalParamMap;
4822                         if ( !$internalParamMap ) {
4823                                 $internalParamMap = array();
4824                                 foreach ( $internalParamNames as $type => $names ) {
4825                                         foreach ( $names as $name ) {
4826                                                 $magicName = str_replace( '-', '_', "img_$name" );
4827                                                 $internalParamMap[$magicName] = array( $type, $name );
4828                                         }
4829                                 }
4830                         }
4831
4832                         // Add handler params
4833                         $paramMap = $internalParamMap;
4834                         if ( $handler ) {
4835                                 $handlerParamMap = $handler->getParamMap();
4836                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4837                                         $paramMap[$magic] = array( 'handler', $paramName );
4838                                 }
4839                         }
4840                         $this->mImageParams[$handlerClass] = $paramMap;
4841                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4842                 }
4843                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4844         }
4845
4846         /**
4847          * Parse image options text and use it to make an image
4848          */
4849         function makeImage( $title, $options ) {
4850                 # @TODO: let the MediaHandler specify its transform parameters
4851                 #
4852                 # Check if the options text is of the form "options|alt text"
4853                 # Options are:
4854                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4855                 #  * left               no resizing, just left align. label is used for alt= only
4856                 #  * right              same, but right aligned
4857                 #  * none               same, but not aligned
4858                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4859                 #  * center             center the image
4860                 #  * framed             Keep original image size, no magnify-button.
4861                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4862                 #  * upright            reduce width for upright images, rounded to full __0 px
4863                 #  * border             draw a 1px border around the image
4864                 # vertical-align values (no % or length right now):
4865                 #  * baseline
4866                 #  * sub
4867                 #  * super
4868                 #  * top
4869                 #  * text-top
4870                 #  * middle
4871                 #  * bottom
4872                 #  * text-bottom
4873
4874                 $parts = array_map( 'trim', explode( '|', $options) );
4875                 $sk = $this->mOptions->getSkin();
4876
4877                 # Give extensions a chance to select the file revision for us
4878                 $skip = $time = false;
4879                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4880
4881                 if ( $skip ) {
4882                         return $sk->makeLinkObj( $title );
4883                 }
4884
4885                 # Get parameter map
4886                 $file = wfFindFile( $title, $time );
4887                 $handler = $file ? $file->getHandler() : false;
4888
4889                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4890
4891                 # Process the input parameters
4892                 $caption = '';
4893                 $params = array( 'frame' => array(), 'handler' => array(),
4894                         'horizAlign' => array(), 'vertAlign' => array() );
4895                 foreach( $parts as $part ) {
4896                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4897                         if ( isset( $paramMap[$magicName] ) ) {
4898                                 list( $type, $paramName ) = $paramMap[$magicName];
4899                                 $params[$type][$paramName] = $value;
4900
4901                                 // Special case; width and height come in one variable together
4902                                 if( $type == 'handler' && $paramName == 'width' ) {
4903                                         $m = array();
4904                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4905                                                 $params[$type]['width'] = intval( $m[1] );
4906                                                 $params[$type]['height'] = intval( $m[2] );
4907                                         } else {
4908                                                 $params[$type]['width'] = intval( $value );
4909                                         }
4910                                 }
4911                         } else {
4912                                 $caption = $part;
4913                         }
4914                 }
4915
4916                 # Process alignment parameters
4917                 if ( $params['horizAlign'] ) {
4918                         $params['frame']['align'] = key( $params['horizAlign'] );
4919                 }
4920                 if ( $params['vertAlign'] ) {
4921                         $params['frame']['valign'] = key( $params['vertAlign'] );
4922                 }
4923
4924                 # Validate the handler parameters
4925                 if ( $handler ) {
4926                         foreach ( $params['handler'] as $name => $value ) {
4927                                 if ( !$handler->validateParam( $name, $value ) ) {
4928                                         unset( $params['handler'][$name] );
4929                                 }
4930                         }
4931                 }
4932
4933                 # Strip bad stuff out of the alt text
4934                 $alt = $this->replaceLinkHoldersText( $caption );
4935
4936                 # make sure there are no placeholders in thumbnail attributes
4937                 # that are later expanded to html- so expand them now and
4938                 # remove the tags
4939                 $alt = $this->mStripState->unstripBoth( $alt );
4940                 $alt = Sanitizer::stripAllTags( $alt );
4941
4942                 $params['frame']['alt'] = $alt;
4943                 $params['frame']['caption'] = $caption;
4944
4945                 # Linker does the rest
4946                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4947
4948                 # Give the handler a chance to modify the parser object
4949                 if ( $handler ) {
4950                         $handler->parserTransformHook( $this, $file );
4951                 }
4952
4953                 return $ret;
4954         }
4955
4956         /**
4957          * Set a flag in the output object indicating that the content is dynamic and
4958          * shouldn't be cached.
4959          */
4960         function disableCache() {
4961                 wfDebug( "Parser output marked as uncacheable.\n" );
4962                 $this->mOutput->mCacheTime = -1;
4963         }
4964
4965         /**#@+
4966          * Callback from the Sanitizer for expanding items found in HTML attribute
4967          * values, so they can be safely tested and escaped.
4968          * @param string $text
4969          * @param PPFrame $frame
4970          * @return string
4971          * @private
4972          */
4973         function attributeStripCallback( &$text, $frame = false ) {
4974                 $text = $this->replaceVariables( $text, $frame );
4975                 $text = $this->mStripState->unstripBoth( $text );
4976                 return $text;
4977         }
4978
4979         /**#@-*/
4980
4981         /**#@+
4982          * Accessor/mutator
4983          */
4984         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4985         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4986         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4987         /**#@-*/
4988
4989         /**#@+
4990          * Accessor
4991          */
4992         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4993         /**#@-*/
4994
4995
4996         /**
4997          * Break wikitext input into sections, and either pull or replace
4998          * some particular section's text.
4999          *
5000          * External callers should use the getSection and replaceSection methods.
5001          *
5002          * @param string $text Page wikitext
5003          * @param string $section A section identifier string of the form:
5004          *   <flag1> - <flag2> - ... - <section number>
5005          *
5006          * Currently the only recognised flag is "T", which means the target section number
5007          * was derived during a template inclusion parse, in other words this is a template
5008          * section edit link. If no flags are given, it was an ordinary section edit link.
5009          * This flag is required to avoid a section numbering mismatch when a section is
5010          * enclosed by <includeonly> (bug 6563).
5011          *
5012          * The section number 0 pulls the text before the first heading; other numbers will
5013          * pull the given section along with its lower-level subsections. If the section is
5014          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5015          *
5016          * @param string $mode One of "get" or "replace"
5017          * @param string $newText Replacement text for section data.
5018          * @return string for "get", the extracted section text.
5019          *                for "replace", the whole page with the section replaced.
5020          */
5021         private function extractSections( $text, $section, $mode, $newText='' ) {
5022                 global $wgTitle;
5023                 $this->clearState();
5024                 $this->mTitle = $wgTitle; // not generally used but removes an ugly failure mode
5025                 $this->mOptions = new ParserOptions;
5026                 $this->setOutputType( OT_WIKI );
5027                 $curIndex = 0;
5028                 $outText = '';
5029                 $frame = new PPFrame( $this );
5030
5031                 // Process section extraction flags
5032                 $flags = 0;
5033                 $sectionParts = explode( '-', $section );
5034                 $sectionIndex = array_pop( $sectionParts );
5035                 foreach ( $sectionParts as $part ) {
5036                         if ( $part == 'T' ) {
5037                                 $flags |= self::PTD_FOR_INCLUSION;
5038                         }
5039                 }
5040                 // Preprocess the text
5041                 $dom = $this->preprocessToDom( $text, $flags );
5042                 $root = $dom->documentElement;
5043
5044                 // <h> nodes indicate section breaks
5045                 // They can only occur at the top level, so we can find them by iterating the root's children
5046                 $node = $root->firstChild;
5047
5048                 // Find the target section
5049                 if ( $sectionIndex == 0 ) {
5050                         // Section zero doesn't nest, level=big
5051                         $targetLevel = 1000;
5052                 } else {
5053                         while ( $node ) {
5054                                 if ( $node->nodeName == 'h' ) {
5055                                         if ( $curIndex + 1 == $sectionIndex ) {
5056                                                 break;
5057                                         }
5058                                         $curIndex++;
5059                                 }
5060                                 if ( $mode == 'replace' ) {
5061                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5062                                 }
5063                                 $node = $node->nextSibling;
5064                         }
5065                         if ( $node ) {
5066                                 $targetLevel = $node->getAttribute( 'level' );
5067                         }
5068                 }
5069
5070                 if ( !$node ) {
5071                         // Not found
5072                         if ( $mode == 'get' ) {
5073                                 return $newText;
5074                         } else {
5075                                 return $text;
5076                         }
5077                 }
5078
5079                 // Find the end of the section, including nested sections
5080                 do {
5081                         if ( $node->nodeName == 'h' ) {
5082                                 $curIndex++;
5083                                 $curLevel = $node->getAttribute( 'level' );
5084                                 if ( $curIndex != $sectionIndex && $curLevel <= $targetLevel ) {
5085                                         break;
5086                                 }
5087                         }
5088                         if ( $mode == 'get' ) {
5089                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5090                         }
5091                         $node = $node->nextSibling;
5092                 } while ( $node );
5093
5094                 // Write out the remainder (in replace mode only)
5095                 if ( $mode == 'replace' ) {
5096                         // Output the replacement text
5097                         // Add two newlines on -- trailing whitespace in $newText is conventionally
5098                         // stripped by the editor, so we need both newlines to restore the paragraph gap
5099                         $outText .= $newText . "\n\n";
5100                         while ( $node ) {
5101                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5102                                 $node = $node->nextSibling;
5103                         }
5104                 }
5105
5106                 if ( is_string( $outText ) ) {
5107                         // Re-insert stripped tags
5108                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5109                 }
5110
5111                 return $outText;
5112         }
5113
5114         /**
5115          * This function returns the text of a section, specified by a number ($section).
5116          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5117          * the first section before any such heading (section 0).
5118          *
5119          * If a section contains subsections, these are also returned.
5120          *
5121          * @param string $text text to look in
5122          * @param string $section section identifier
5123          * @param string $deftext default to return if section is not found
5124          * @return string text of the requested section
5125          */
5126         public function getSection( $text, $section, $deftext='' ) {
5127                 return $this->extractSections( $text, $section, "get", $deftext );
5128         }
5129
5130         public function replaceSection( $oldtext, $section, $text ) {
5131                 return $this->extractSections( $oldtext, $section, "replace", $text );
5132         }
5133
5134         /**
5135          * Get the timestamp associated with the current revision, adjusted for
5136          * the default server-local timestamp
5137          */
5138         function getRevisionTimestamp() {
5139                 if ( is_null( $this->mRevisionTimestamp ) ) {
5140                         wfProfileIn( __METHOD__ );
5141                         global $wgContLang;
5142                         $dbr = wfGetDB( DB_SLAVE );
5143                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5144                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5145
5146                         // Normalize timestamp to internal MW format for timezone processing.
5147                         // This has the added side-effect of replacing a null value with
5148                         // the current time, which gives us more sensible behavior for
5149                         // previews.
5150                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5151
5152                         // The cryptic '' timezone parameter tells to use the site-default
5153                         // timezone offset instead of the user settings.
5154                         //
5155                         // Since this value will be saved into the parser cache, served
5156                         // to other users, and potentially even used inside links and such,
5157                         // it needs to be consistent for all visitors.
5158                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5159
5160                         wfProfileOut( __METHOD__ );
5161                 }
5162                 return $this->mRevisionTimestamp;
5163         }
5164
5165         /**
5166          * Mutator for $mDefaultSort
5167          *
5168          * @param $sort New value
5169          */
5170         public function setDefaultSort( $sort ) {
5171                 $this->mDefaultSort = $sort;
5172         }
5173
5174         /**
5175          * Accessor for $mDefaultSort
5176          * Will use the title/prefixed title if none is set
5177          *
5178          * @return string
5179          */
5180         public function getDefaultSort() {
5181                 if( $this->mDefaultSort !== false ) {
5182                         return $this->mDefaultSort;
5183                 } else {
5184                         return $this->mTitle->getNamespace() == NS_CATEGORY
5185                                         ? $this->mTitle->getText()
5186                                         : $this->mTitle->getPrefixedText();
5187                 }
5188         }
5189
5190         /**
5191          * Try to guess the section anchor name based on a wikitext fragment
5192          * presumably extracted from a heading, for example "Header" from
5193          * "== Header ==".
5194          */
5195         public function guessSectionNameFromWikiText( $text ) {
5196                 # Strip out wikitext links(they break the anchor)
5197                 $text = $this->stripSectionName( $text );
5198                 $headline = Sanitizer::decodeCharReferences( $text );
5199                 # strip out HTML
5200                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5201                 $headline = trim( $headline );
5202                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5203                 $replacearray = array(
5204                         '%3A' => ':',
5205                         '%' => '.'
5206                 );
5207                 return str_replace(
5208                         array_keys( $replacearray ),
5209                         array_values( $replacearray ),
5210                         $sectionanchor );
5211         }
5212
5213         /**
5214          * Strips a text string of wikitext for use in a section anchor
5215          *
5216          * Accepts a text string and then removes all wikitext from the
5217          * string and leaves only the resultant text (i.e. the result of
5218          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5219          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5220          * to create valid section anchors by mimicing the output of the
5221          * parser when headings are parsed.
5222          *
5223          * @param $text string Text string to be stripped of wikitext
5224          * for use in a Section anchor
5225          * @return Filtered text string
5226          */
5227         public function stripSectionName( $text ) {
5228                 # Strip internal link markup
5229                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5230                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5231
5232                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5233                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5234                 # on how many empty links there are on the page - need to figure that out.
5235                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5236
5237                 # Parse wikitext quotes (italics & bold)
5238                 $text = $this->doQuotes($text);
5239
5240                 # Strip HTML tags
5241                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5242                 return $text;
5243         }
5244
5245         /**
5246          * strip/replaceVariables/unstrip for preprocessor regression testing
5247          */
5248         function srvus( $text ) {
5249                 $text = $this->replaceVariables( $text );
5250                 $text = $this->mStripState->unstripBoth( $text );
5251                 $text = Sanitizer::removeHTMLtags( $text );
5252                 return $text;
5253         }
5254 }
5255
5256 /**
5257  * @todo document, briefly.
5258  * @addtogroup Parser
5259  */
5260 class OnlyIncludeReplacer {
5261         var $output = '';
5262
5263         function replace( $matches ) {
5264                 if ( substr( $matches[1], -1 ) == "\n" ) {
5265                         $this->output .= substr( $matches[1], 0, -1 );
5266                 } else {
5267                         $this->output .= $matches[1];
5268                 }
5269         }
5270 }
5271
5272 /**
5273  * @todo document, briefly.
5274  * @addtogroup Parser
5275  */
5276 class StripState {
5277         var $general, $nowiki;
5278
5279         function __construct() {
5280                 $this->general = new ReplacementArray;
5281                 $this->nowiki = new ReplacementArray;
5282         }
5283
5284         function unstripGeneral( $text ) {
5285                 wfProfileIn( __METHOD__ );
5286                 do {
5287                         $oldText = $text;
5288                         $text = $this->general->replace( $text );
5289                 } while ( $text != $oldText );
5290                 wfProfileOut( __METHOD__ );
5291                 return $text;
5292         }
5293
5294         function unstripNoWiki( $text ) {
5295                 wfProfileIn( __METHOD__ );
5296                 do {
5297                         $oldText = $text;
5298                         $text = $this->nowiki->replace( $text );
5299                 } while ( $text != $oldText );
5300                 wfProfileOut( __METHOD__ );
5301                 return $text;
5302         }
5303
5304         function unstripBoth( $text ) {
5305                 wfProfileIn( __METHOD__ );
5306                 do {
5307                         $oldText = $text;
5308                         $text = $this->general->replace( $text );
5309                         $text = $this->nowiki->replace( $text );
5310                 } while ( $text != $oldText );
5311                 wfProfileOut( __METHOD__ );
5312                 return $text;
5313         }
5314 }
5315
5316 /**
5317  * An expansion frame, used as a context to expand the result of preprocessToDom()
5318  */
5319 class PPFrame {
5320         var $parser, $title;
5321         var $titleCache;
5322
5323         /**
5324          * Hashtable listing templates which are disallowed for expansion in this frame,
5325          * having been encountered previously in parent frames.
5326          */
5327         var $loopCheckHash;
5328
5329         /**
5330          * Recursion depth of this frame, top = 0
5331          */
5332         var $depth;
5333
5334         const NO_ARGS = 1;
5335         const NO_TEMPLATES = 2;
5336         const STRIP_COMMENTS = 4;
5337         const NO_IGNORE = 8;
5338
5339         const RECOVER_ORIG = 11;
5340
5341         /**
5342          * Construct a new preprocessor frame.
5343          * @param Parser $parser The parent parser
5344          * @param Title $title The context title, or false if there isn't one
5345          */
5346         function __construct( $parser ) {
5347                 $this->parser = $parser;
5348                 $this->title = $parser->mTitle;
5349                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
5350                 $this->loopCheckHash = array();
5351                 $this->depth = 0;
5352         }
5353
5354         /**
5355          * Create a new child frame
5356          * $args is optionally a DOMNodeList containing the template arguments
5357          */
5358         function newChild( $args = false, $title = false ) {
5359                 $namedArgs = array();
5360                 $numberedArgs = array();
5361                 if ( $title === false ) {
5362                         $title = $this->title;
5363                 }
5364                 if ( $args !== false ) {
5365                         $xpath = false;
5366                         foreach ( $args as $arg ) {
5367                                 if ( !$xpath ) {
5368                                         $xpath = new DOMXPath( $arg->ownerDocument );
5369                                 }
5370
5371                                 $nameNodes = $xpath->query( 'name', $arg );
5372                                 $value = $xpath->query( 'value', $arg );
5373                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5374                                         // Numbered parameter
5375                                         $index = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5376                                         $numberedArgs[$index] = $value->item( 0 );
5377                                         unset( $namedArgs[$index] );
5378                                 } else {
5379                                         // Named parameter
5380                                         $name = trim( $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS ) );
5381                                         $namedArgs[$name] = $value->item( 0 );
5382                                         unset( $numberedArgs[$name] );
5383                                 }
5384                         }
5385                 }
5386                 return new PPTemplateFrame( $this->parser, $this, $numberedArgs, $namedArgs, $title );
5387         }
5388
5389         /**
5390          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5391          * using the current context
5392          * @param $root the node
5393          */
5394         function expand( $root, $flags = 0 ) {
5395                 if ( is_string( $root ) ) {
5396                         return $root;
5397                 }
5398
5399                 if ( $this->parser->ot['html']
5400                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5401                 {
5402                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5403                 }
5404
5405                 if ( is_array( $root ) ) {
5406                         $s = '';
5407                         foreach ( $root as $node ) {
5408                                 $s .= $this->expand( $node, $flags );
5409                         }
5410                 } elseif ( $root instanceof DOMNodeList ) {
5411                         $s = '';
5412                         foreach ( $root as $node ) {
5413                                 $s .= $this->expand( $node, $flags );
5414                         }
5415                 } elseif ( $root instanceof DOMNode ) {
5416                         if ( $root->nodeType == XML_TEXT_NODE ) {
5417                                 $s = $root->nodeValue;
5418                         } elseif ( $root->nodeName == 'template' ) {
5419                                 # Double-brace expansion
5420                                 $xpath = new DOMXPath( $root->ownerDocument );
5421                                 $titles = $xpath->query( 'title', $root );
5422                                 $title = $titles->item( 0 );
5423                                 $parts = $xpath->query( 'part', $root );
5424                                 if ( $flags & self::NO_TEMPLATES ) {
5425                                         $s = '{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}';
5426                                 } else {
5427                                         $lineStart = $root->getAttribute( 'lineStart' );
5428                                         $params = array(
5429                                                 'title' => $title,
5430                                                 'parts' => $parts,
5431                                                 'lineStart' => $lineStart,
5432                                                 'text' => 'FIXME' );
5433                                         $s = $this->parser->braceSubstitution( $params, $this );
5434                                 }
5435                         } elseif ( $root->nodeName == 'tplarg' ) {
5436                                 # Triple-brace expansion
5437                                 $xpath = new DOMXPath( $root->ownerDocument );
5438                                 $titles = $xpath->query( 'title', $root );
5439                                 $title = $titles->item( 0 );
5440                                 $parts = $xpath->query( 'part', $root );
5441                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5442                                         $s = '{{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}}';
5443                                 } else {
5444                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5445                                         $s = $this->parser->argSubstitution( $params, $this );
5446                                 }
5447                         } elseif ( $root->nodeName == 'comment' ) {
5448                                 # HTML-style comment
5449                                 if ( $this->parser->ot['html']
5450                                         || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
5451                                         || ( $flags & self::STRIP_COMMENTS ) )
5452                                 {
5453                                         $s = '';
5454                                 } else {
5455                                         $s = $root->textContent;
5456                                 }
5457                         } elseif ( $root->nodeName == 'ignore' ) {
5458                                 # Output suppression used by <includeonly> etc.
5459                                 # OT_WIKI will only respect <ignore> in substed templates.
5460                                 # The other output types respect it unless NO_IGNORE is set.
5461                                 # extractSections() sets NO_IGNORE and so never respects it.
5462                                 if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & self::NO_IGNORE ) ) {
5463                                         $s = $root->textContent;
5464                                 } else {
5465                                         $s = '';
5466                                 }
5467                         } elseif ( $root->nodeName == 'ext' ) {
5468                                 # Extension tag
5469                                 $xpath = new DOMXPath( $root->ownerDocument );
5470                                 $names = $xpath->query( 'name', $root );
5471                                 $attrs = $xpath->query( 'attr', $root );
5472                                 $inners = $xpath->query( 'inner', $root );
5473                                 $closes = $xpath->query( 'close', $root );
5474                                 $params = array(
5475                                         'name' => $names->item( 0 ),
5476                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5477                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5478                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5479                                 );
5480                                 $s = $this->parser->extensionSubstitution( $params, $this );
5481                         } elseif ( $root->nodeName == 'h' ) {
5482                                 # Heading
5483                                 $s = $this->expand( $root->childNodes, $flags );
5484
5485                                 if ( $this->parser->ot['html'] ) {
5486                                         # Insert heading index marker
5487                                         $headingIndex = $root->getAttribute( 'i' );
5488                                         $titleText = $this->title->getPrefixedDBkey();
5489                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5490                                         $serial = count( $this->parser->mHeadings ) - 1;
5491                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5492                                         $count = $root->getAttribute( 'level' );
5493                                         $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5494                                         $this->parser->mStripState->general->setPair( $marker, '' );
5495                                 }
5496                         } else {
5497                                 # Generic recursive expansion
5498                                 $s = '';
5499                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5500                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5501                                                 $s .= $node->nodeValue;
5502                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5503                                                 $s .= $this->expand( $node, $flags );
5504                                         }
5505                                 }
5506                         }
5507                 } else {
5508                         throw new MWException( __METHOD__.': Invalid parameter type' );
5509                 }
5510                 return $s;
5511         }
5512
5513         function implodeWithFlags( $sep, $flags /*, ... */ ) {
5514                 $args = array_slice( func_get_args(), 2 );
5515
5516                 $first = true;
5517                 $s = '';
5518                 foreach ( $args as $root ) {
5519                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5520                                 $root = array( $root );
5521                         }
5522                         foreach ( $root as $node ) {
5523                                 if ( $first ) {
5524                                         $first = false;
5525                                 } else {
5526                                         $s .= $sep;
5527                                 }
5528                                 $s .= $this->expand( $node, $flags );
5529                         }
5530                 }
5531                 return $s;
5532         }
5533
5534         function implode( $sep /*, ... */ ) {
5535                 $args = func_get_args();
5536                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0 ), array_slice( $args, 1 ) );
5537                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5538         }
5539
5540         /**
5541          * Split an <arg> or <template> node into a three-element array:
5542          *    DOMNode name, string index and DOMNode value
5543          */
5544         function splitBraceNode( $node ) {
5545                 $xpath = new DOMXPath( $node->ownerDocument );
5546                 $names = $xpath->query( 'name', $node );
5547                 $values = $xpath->query( 'value', $node );
5548                 if ( !$names->length || !$values->length ) {
5549                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
5550                 }
5551                 $name = $names->item( 0 );
5552                 $index = $name->getAttribute( 'index' );
5553                 return array( $name, $index, $values->item( 0 ) );
5554         }
5555
5556         /**
5557          * Split an <ext> node into an associative array containing name, attr, inner and close
5558          * All values in the resulting array are DOMNodes. Inner and close are optional.
5559          */
5560         function splitExtNode( $node ) {
5561                 $xpath = new DOMXPath( $node->ownerDocument );
5562                 $names = $xpath->query( 'name', $node );
5563                 $attrs = $xpath->query( 'attr', $node );
5564                 $inners = $xpath->query( 'inner', $node );
5565                 $closes = $xpath->query( 'close', $node );
5566                 if ( !$names->length || !$attrs->length ) {
5567                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
5568                 }
5569                 $parts = array(
5570                         'name' => $names->item( 0 ),
5571                         'attr' => $attrs->item( 0 ) );
5572                 if ( $inners->length ) {
5573                         $parts['inner'] = $inners->item( 0 );
5574                 }
5575                 if ( $closes->length ) {
5576                         $parts['close'] = $closes->item( 0 );
5577                 }
5578                 return $parts;
5579         }
5580
5581         function __toString() {
5582                 return 'frame{}';
5583         }
5584
5585         function getPDBK( $level = false ) {
5586                 if ( $level === false ) {
5587                         return $this->title->getPrefixedDBkey();
5588                 } else {
5589                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
5590                 }
5591         }
5592
5593         /**
5594          * Returns true if there are no arguments in this frame
5595          */
5596         function isEmpty() {
5597                 return true;
5598         }
5599
5600         function getArgument( $name ) {
5601                 return false;
5602         }
5603 }
5604
5605 /**
5606  * Expansion frame with template arguments
5607  */
5608 class PPTemplateFrame extends PPFrame {
5609         var $numberedArgs, $namedArgs, $parent;
5610         var $numberedExpansionCache, $namedExpansionCache;
5611
5612         function __construct( $parser, $parent = false, $numberedArgs = array(), $namedArgs = array(), $title = false ) {
5613                 $this->parser = $parser;
5614                 $this->parent = $parent;
5615                 $this->numberedArgs = $numberedArgs;
5616                 $this->namedArgs = $namedArgs;
5617                 $this->title = $title;
5618                 $this->titleCache = $parent->titleCache;
5619                 $this->titleCache[] = $title ? $title->getPrefixedDBkey() : false;
5620                 $this->loopCheckHash = /*clone*/ $parent->loopCheckHash;
5621                 $this->depth = $parent->depth + 1;
5622                 $this->numberedExpansionCache = $this->namedExpansionCache = array();
5623         }
5624
5625         function __toString() {
5626                 $s = 'tplframe{';
5627                 $first = true;
5628                 $args = $this->numberedArgs + $this->namedArgs;
5629                 foreach ( $args as $name => $value ) {
5630                         if ( $first ) {
5631                                 $first = false;
5632                         } else {
5633                                 $s .= ', ';
5634                         }
5635                         $s .= "\"$name\":\"" .
5636                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5637                 }
5638                 $s .= '}';
5639                 return $s;
5640         }
5641         /**
5642          * Returns true if there are no arguments in this frame
5643          */
5644         function isEmpty() {
5645                 return !count( $this->numberedArgs ) && !count( $this->namedArgs );
5646         }
5647
5648         function getNumberedArgument( $index ) {
5649                 if ( !isset( $this->numberedArgs[$index] ) ) {
5650                         return false;
5651                 }
5652                 if ( !isset( $this->numberedExpansionCache[$index] ) ) {
5653                         # No trimming for unnamed arguments
5654                         $this->numberedExpansionCache[$index] = $this->parent->expand( $this->numberedArgs[$index], self::STRIP_COMMENTS );
5655                 }
5656                 return $this->numberedExpansionCache[$index];
5657         }
5658
5659         function getNamedArgument( $name ) {
5660                 if ( !isset( $this->namedArgs[$name] ) ) {
5661                         return false;
5662                 }
5663                 if ( !isset( $this->namedExpansionCache[$name] ) ) {
5664                         # Trim named arguments post-expand, for backwards compatibility
5665                         $this->namedExpansionCache[$name] = trim(
5666                                 $this->parent->expand( $this->namedArgs[$name], self::STRIP_COMMENTS ) );
5667                 }
5668                 return $this->namedExpansionCache[$name];
5669         }
5670
5671         function getArgument( $name ) {
5672                 wfDebug( __METHOD__." getting '$name'\n" );
5673                 $text = $this->getNumberedArgument( $name );
5674                 if ( $text === false ) {
5675                         $text = $this->getNamedArgument( $name );
5676                 }
5677                 return $text;
5678         }
5679 }