includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10 /**
  11  * Update this version number when the ParserOutput format
  12  * changes in an incompatible way, so the parser cache
  13  * can automatically discard old data.
  14  */
  15 define( 'MW_PARSER_VERSION', '1.6.2' );
  16
  17 define( 'RLH_FOR_UPDATE', 1 );
  18
  19 # Allowed values for $mOutputType
  20 define( 'OT_HTML', 1 );
  21 define( 'OT_WIKI', 2 );
  22 define( 'OT_MSG' , 3 );
  23 define( 'OT_PREPROCESS', 4 );
  24
  25 # Flags for setFunctionHook
  26 define( 'SFH_NO_HASH', 1 );
  27
  28 # string parameter for extractTags which will cause it
  29 # to strip HTML comments in addition to regular
  30 # <XML>-style tags. This should not be anything we
  31 # may want to use in wikisyntax
  32 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  33
  34 # Constants needed for external link processing
  35 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  36 # Everything except bracket, space, or control characters
  37 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  38 # Including space, but excluding newlines
  39 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  40 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  41 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  42 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  43         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  44 define( 'EXT_IMAGE_REGEX',
  45         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  46         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  47         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  48 );
  49
  50 // State constants for the definition list colon extraction
  51 define( 'MW_COLON_STATE_TEXT', 0 );
  52 define( 'MW_COLON_STATE_TAG', 1 );
  53 define( 'MW_COLON_STATE_TAGSTART', 2 );
  54 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  55 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  56 define( 'MW_COLON_STATE_COMMENT', 5 );
  57 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  58 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  59
  60 /**
  61  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  62  * syntax, such as "[[link]]" for making links), and provides a one-way
  63  * transformation of that wiki markup it into XHTML output / markup
  64  * (which in turn the browser understands, and can display).
  65  *
  66  * <pre>
  67  * There are four main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  * preprocess()
  75  *   removes HTML comments and expands templates
  76  *
  77  * Globals used:
  78  *    objects:   $wgLang, $wgContLang
  79  *
  80  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  81  *
  82  * settings:
  83  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  84  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  85  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  86  *  $wgMaxArticleSize*
  87  *
  88  *  * only within ParserOptions
  89  * </pre>
  90  *
  91  * @addtogroup Parser
  92  */
  93 class Parser
  94 {
  95         const VERSION = MW_PARSER_VERSION;
  96         /**#@+
  97          * @private
  98          */
  99         # Persistent:
 100         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
 101
 102         # Cleared with clearState():
 103         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 104         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 105         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 106         var $mIncludeSizes, $mDefaultSort;
 107         var $mTemplates,        // cache of already loaded templates, avoids
 108                                 // multiple SQL queries for the same string
 109             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 110                                 // in this path. Used for loop detection.
 111
 112         # Temporary
 113         # These are variables reset at least once per parse regardless of $clearState
 114         var $mOptions,      // ParserOptions object
 115                 $mTitle,        // Title context, used for self-link rendering and similar things
 116                 $mOutputType,   // Output type, one of the OT_xxx constants
 117                 $ot,            // Shortcut alias, see setOutputType()
 118                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 119                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 120                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 121
 122         /**#@-*/
 123
 124         /**
 125          * Constructor
 126          *
 127          * @public
 128          */
 129         function Parser() {
 130                 $this->mTagHooks = array();
 131                 $this->mTransparentTagHooks = array();
 132                 $this->mFunctionHooks = array();
 133                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 134                 $this->mFirstCall = true;
 135         }
 136
 137         /**
 138          * Do various kinds of initialisation on the first call of the parser
 139          */
 140         function firstCallInit() {
 141                 if ( !$this->mFirstCall ) {
 142                         return;
 143                 }
 144
 145                 wfProfileIn( __METHOD__ );
 146                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 147
 148                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 149
 150                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 170                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 171                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 172                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 173                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 174                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 175                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 176
 177                 if ( $wgAllowDisplayTitle ) {
 178                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 179                 }
 180                 if ( $wgAllowSlowParserFunctions ) {
 181                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 182                 }
 183
 184                 $this->initialiseVariables();
 185                 $this->mFirstCall = false;
 186                 wfProfileOut( __METHOD__ );
 187         }
 188
 189         /**
 190          * Clear Parser state
 191          *
 192          * @private
 193          */
 194         function clearState() {
 195                 wfProfileIn( __METHOD__ );
 196                 if ( $this->mFirstCall ) {
 197                         $this->firstCallInit();
 198                 }
 199                 $this->mOutput = new ParserOutput;
 200                 $this->mAutonumber = 0;
 201                 $this->mLastSection = '';
 202                 $this->mDTopen = false;
 203                 $this->mIncludeCount = array();
 204                 $this->mStripState = new StripState;
 205                 $this->mArgStack = array();
 206                 $this->mInPre = false;
 207                 $this->mInterwikiLinkHolders = array(
 208                         'texts' => array(),
 209                         'titles' => array()
 210                 );
 211                 $this->mLinkHolders = array(
 212                         'namespaces' => array(),
 213                         'dbkeys' => array(),
 214                         'queries' => array(),
 215                         'texts' => array(),
 216                         'titles' => array()
 217                 );
 218                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 219
 220                 /**
 221                  * Prefix for temporary replacement strings for the multipass parser.
 222                  * \x07 should never appear in input as it's disallowed in XML.
 223                  * Using it at the front also gives us a little extra robustness
 224                  * since it shouldn't match when butted up against identifier-like
 225                  * string constructs.
 226                  */
 227                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 228
 229                 # Clear these on every parse, bug 4549
 230                 $this->mTemplates = array();
 231                 $this->mTemplatePath = array();
 232
 233                 $this->mShowToc = true;
 234                 $this->mForceTocPosition = false;
 235                 $this->mIncludeSizes = array(
 236                         'pre-expand' => 0,
 237                         'post-expand' => 0,
 238                         'arg' => 0
 239                 );
 240                 $this->mDefaultSort = false;
 241
 242                 wfRunHooks( 'ParserClearState', array( &$this ) );
 243                 wfProfileOut( __METHOD__ );
 244         }
 245
 246         function setOutputType( $ot ) {
 247                 $this->mOutputType = $ot;
 248                 // Shortcut alias
 249                 $this->ot = array(
 250                         'html' => $ot == OT_HTML,
 251                         'wiki' => $ot == OT_WIKI,
 252                         'msg' => $ot == OT_MSG,
 253                         'pre' => $ot == OT_PREPROCESS,
 254                 );
 255         }
 256
 257         /**
 258          * Accessor for mUniqPrefix.
 259          *
 260          * @public
 261          */
 262         function uniqPrefix() {
 263                 return $this->mUniqPrefix;
 264         }
 265
 266         /**
 267          * Convert wikitext to HTML
 268          * Do not call this function recursively.
 269          *
 270          * @param string $text Text we want to parse
 271          * @param Title &$title A title object
 272          * @param array $options
 273          * @param boolean $linestart
 274          * @param boolean $clearState
 275          * @param int $revid number to pass in {{REVISIONID}}
 276          * @return ParserOutput a ParserOutput
 277          */
 278         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 279                 /**
 280                  * First pass--just handle <nowiki> sections, pass the rest off
 281                  * to internalParse() which does all the real work.
 282                  */
 283
 284                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 285                 $fname = 'Parser::parse-' . wfGetCaller();
 286                 wfProfileIn( __METHOD__ );
 287                 wfProfileIn( $fname );
 288
 289                 if ( $clearState ) {
 290                         $this->clearState();
 291                 }
 292
 293                 $this->mOptions = $options;
 294                 $this->mTitle =& $title;
 295                 $oldRevisionId = $this->mRevisionId;
 296                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 297                 if( $revid !== null ) {
 298                         $this->mRevisionId = $revid;
 299                         $this->mRevisionTimestamp = null;
 300                 }
 301                 $this->setOutputType( OT_HTML );
 302                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 303                 $text = $this->strip( $text, $this->mStripState );
 304                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 305                 $text = $this->internalParse( $text );
 306                 $text = $this->mStripState->unstripGeneral( $text );
 307
 308                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 309                 $fixtags = array(
 310                         # french spaces, last one Guillemet-left
 311                         # only if there is something before the space
 312                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 313                         # french spaces, Guillemet-right
 314                         '/(\\302\\253) /' => '\\1&nbsp;',
 315                 );
 316                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 317
 318                 # only once and last
 319                 $text = $this->doBlockLevels( $text, $linestart );
 320
 321                 $this->replaceLinkHolders( $text );
 322
 323                 # the position of the parserConvert() call should not be changed. it
 324                 # assumes that the links are all replaced and the only thing left
 325                 # is the <nowiki> mark.
 326                 # Side-effects: this calls $this->mOutput->setTitleText()
 327                 $text = $wgContLang->parserConvert( $text, $this );
 328
 329                 $text = $this->mStripState->unstripNoWiki( $text );
 330
 331                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 332
 333 //!JF Move to its own function
 334
 335                 $uniq_prefix = $this->mUniqPrefix;
 336                 $matches = array();
 337                 $elements = array_keys( $this->mTransparentTagHooks );
 338                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 339
 340                 foreach( $matches as $marker => $data ) {
 341                         list( $element, $content, $params, $tag ) = $data;
 342                         $tagName = strtolower( $element );
 343                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 344                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 345                                         array( $content, $params, $this ) );
 346                         } else {
 347                                 $output = $tag;
 348                         }
 349                         $this->mStripState->general->setPair( $marker, $output );
 350                 }
 351                 $text = $this->mStripState->unstripGeneral( $text );
 352
 353
 354
 355
 356
 357
 358
 359
 360                 $text = Sanitizer::normalizeCharReferences( $text );
 361
 362                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 363                         $text = Parser::tidy($text);
 364                 } else {
 365                         # attempt to sanitize at least some nesting problems
 366                         # (bug #2702 and quite a few others)
 367                         $tidyregs = array(
 368                                 # ''Something [http://www.cool.com cool''] -->
 369                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 370                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 371                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 372                                 # fix up an anchor inside another anchor, only
 373                                 # at least for a single single nested link (bug 3695)
 374                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 375                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 376                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 377                                 # contains a div, so fix it up here; replace
 378                                 # div with escaped text
 379                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 380                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 381                                 # remove empty italic or bold tag pairs, some
 382                                 # introduced by rules above
 383                                 '/<([bi])><\/\\1>/' => '',
 384                         );
 385
 386                         $text = preg_replace(
 387                                 array_keys( $tidyregs ),
 388                                 array_values( $tidyregs ),
 389                                 $text );
 390                 }
 391
 392                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 393
 394                 # Information on include size limits, for the benefit of users who try to skirt them
 395                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 396                         $max = $this->mOptions->getMaxIncludeSize();
 397                         $text .= "<!-- \n" .
 398                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 399                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 400                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 401                                 "Maximum: $max bytes\n" .
 402                                 "-->\n";
 403                 }
 404                 $this->mOutput->setText( $text );
 405                 $this->mRevisionId = $oldRevisionId;
 406                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 407                 wfProfileOut( $fname );
 408                 wfProfileOut( __METHOD__ );
 409
 410                 return $this->mOutput;
 411         }
 412
 413         /**
 414          * Recursive parser entry point that can be called from an extension tag
 415          * hook.
 416          */
 417         function recursiveTagParse( $text ) {
 418                 wfProfileIn( __METHOD__ );
 419                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 420                 $text = $this->strip( $text, $this->mStripState );
 421                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 422                 $text = $this->internalParse( $text );
 423                 wfProfileOut( __METHOD__ );
 424                 return $text;
 425         }
 426
 427         /**
 428          * Expand templates and variables in the text, producing valid, static wikitext.
 429          * Also removes comments.
 430          */
 431         function preprocess( $text, $title, $options, $revid = null ) {
 432                 wfProfileIn( __METHOD__ );
 433                 $this->clearState();
 434                 $this->setOutputType( OT_PREPROCESS );
 435                 $this->mOptions = $options;
 436                 $this->mTitle = $title;
 437                 if( $revid !== null ) {
 438                         $this->mRevisionId = $revid;
 439                 }
 440                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 441                 $text = $this->strip( $text, $this->mStripState );
 442                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 443                 if ( $this->mOptions->getRemoveComments() ) {
 444                         $text = Sanitizer::removeHTMLcomments( $text );
 445                 }
 446                 $text = $this->replaceVariables( $text );
 447                 $text = $this->mStripState->unstripBoth( $text );
 448                 wfProfileOut( __METHOD__ );
 449                 return $text;
 450         }
 451
 452         /**
 453          * Get a random string
 454          *
 455          * @private
 456          * @static
 457          */
 458         function getRandomString() {
 459                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 460         }
 461
 462         function &getTitle() { return $this->mTitle; }
 463         function getOptions() { return $this->mOptions; }
 464
 465         function getFunctionLang() {
 466                 global $wgLang, $wgContLang;
 467                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 468         }
 469
 470         /**
 471          * Replaces all occurrences of HTML-style comments and the given tags
 472          * in the text with a random marker and returns teh next text. The output
 473          * parameter $matches will be an associative array filled with data in
 474          * the form:
 475          *   'UNIQ-xxxxx' => array(
 476          *     'element',
 477          *     'tag content',
 478          *     array( 'param' => 'x' ),
 479          *     '<element param="x">tag content</element>' ) )
 480          *
 481          * @param $elements list of element names. Comments are always extracted.
 482          * @param $text Source text string.
 483          * @param $uniq_prefix
 484          *
 485          * @public
 486          * @static
 487          */
 488         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 489                 static $n = 1;
 490                 $stripped = '';
 491                 $matches = array();
 492
 493                 $taglist = implode( '|', $elements );
 494                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 495
 496                 while ( '' != $text ) {
 497                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 498                         $stripped .= $p[0];
 499                         if( count( $p ) < 5 ) {
 500                                 break;
 501                         }
 502                         if( count( $p ) > 5 ) {
 503                                 // comment
 504                                 $element    = $p[4];
 505                                 $attributes = '';
 506                                 $close      = '';
 507                                 $inside     = $p[5];
 508                         } else {
 509                                 // tag
 510                                 $element    = $p[1];
 511                                 $attributes = $p[2];
 512                                 $close      = $p[3];
 513                                 $inside     = $p[4];
 514                         }
 515
 516                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 517                         $stripped .= $marker;
 518
 519                         if ( $close === '/>' ) {
 520                                 // Empty element tag, <tag />
 521                                 $content = null;
 522                                 $text = $inside;
 523                                 $tail = null;
 524                         } else {
 525                                 if( $element == '!--' ) {
 526                                         $end = '/(-->)/';
 527                                 } else {
 528                                         $end = "/(<\\/$element\\s*>)/i";
 529                                 }
 530                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 531                                 $content = $q[0];
 532                                 if( count( $q ) < 3 ) {
 533                                         # No end tag -- let it run out to the end of the text.
 534                                         $tail = '';
 535                                         $text = '';
 536                                 } else {
 537                                         $tail = $q[1];
 538                                         $text = $q[2];
 539                                 }
 540                         }
 541
 542                         $matches[$marker] = array( $element,
 543                                 $content,
 544                                 Sanitizer::decodeTagAttributes( $attributes ),
 545                                 "<$element$attributes$close$content$tail" );
 546                 }
 547                 return $stripped;
 548         }
 549
 550         /**
 551          * Strips and renders nowiki, pre, math, hiero
 552          * If $render is set, performs necessary rendering operations on plugins
 553          * Returns the text, and fills an array with data needed in unstrip()
 554          *
 555          * @param StripState $state
 556          *
 557          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 558          *  will be stripped in addition to other tags. This is important
 559          *  for section editing, where these comments cause confusion when
 560          *  counting the sections in the wikisource
 561          *
 562          * @param array dontstrip contains tags which should not be stripped;
 563          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 564          *
 565          * @private
 566          */
 567         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 568                 global $wgContLang;
 569                 wfProfileIn( __METHOD__ );
 570                 $render = ($this->mOutputType == OT_HTML);
 571
 572                 $uniq_prefix = $this->mUniqPrefix;
 573                 $commentState = new ReplacementArray;
 574                 $nowikiItems = array();
 575                 $generalItems = array();
 576
 577                 $elements = array_merge(
 578                         array( 'nowiki', 'gallery' ),
 579                         array_keys( $this->mTagHooks ) );
 580                 global $wgRawHtml;
 581                 if( $wgRawHtml ) {
 582                         $elements[] = 'html';
 583                 }
 584                 if( $this->mOptions->getUseTeX() ) {
 585                         $elements[] = 'math';
 586                 }
 587
 588                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 589                 foreach ( $elements AS $k => $v ) {
 590                         if ( !in_array ( $v , $dontstrip ) ) continue;
 591                         unset ( $elements[$k] );
 592                 }
 593
 594                 $matches = array();
 595                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 596
 597                 foreach( $matches as $marker => $data ) {
 598                         list( $element, $content, $params, $tag ) = $data;
 599                         if( $render ) {
 600                                 $tagName = strtolower( $element );
 601                                 wfProfileIn( __METHOD__."-render-$tagName" );
 602                                 switch( $tagName ) {
 603                                 case '!--':
 604                                         // Comment
 605                                         if( substr( $tag, -3 ) == '-->' ) {
 606                                                 $output = $tag;
 607                                         } else {
 608                                                 // Unclosed comment in input.
 609                                                 // Close it so later stripping can remove it
 610                                                 $output = "$tag-->";
 611                                         }
 612                                         break;
 613                                 case 'html':
 614                                         if( $wgRawHtml ) {
 615                                                 $output = $content;
 616                                                 break;
 617                                         }
 618                                         // Shouldn't happen otherwise. :)
 619                                 case 'nowiki':
 620                                         $output = Xml::escapeTagsOnly( $content );
 621                                         break;
 622                                 case 'math':
 623                                         $output = $wgContLang->armourMath(
 624                                                 MathRenderer::renderMath( $content, $params ) );
 625                                         break;
 626                                 case 'gallery':
 627                                         $output = $this->renderImageGallery( $content, $params );
 628                                         break;
 629                                 default:
 630                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 631                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 632                                                         array( $content, $params, $this ) );
 633                                         } else {
 634                                                 throw new MWException( "Invalid call hook $element" );
 635                                         }
 636                                 }
 637                                 wfProfileOut( __METHOD__."-render-$tagName" );
 638                         } else {
 639                                 // Just stripping tags; keep the source
 640                                 $output = $tag;
 641                         }
 642
 643                         // Unstrip the output, to support recursive strip() calls
 644                         $output = $state->unstripBoth( $output );
 645
 646                         if( !$stripcomments && $element == '!--' ) {
 647                                 $commentState->setPair( $marker, $output );
 648                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 649                                 $nowikiItems[$marker] = $output;
 650                         } else {
 651                                 $generalItems[$marker] = $output;
 652                         }
 653                 }
 654                 # Add the new items to the state
 655                 # We do this after the loop instead of during it to avoid slowing
 656                 # down the recursive unstrip
 657                 $state->nowiki->mergeArray( $nowikiItems );
 658                 $state->general->mergeArray( $generalItems );
 659
 660                 # Unstrip comments unless explicitly told otherwise.
 661                 # (The comments are always stripped prior to this point, so as to
 662                 # not invoke any extension tags / parser hooks contained within
 663                 # a comment.)
 664                 if ( !$stripcomments ) {
 665                         // Put them all back and forget them
 666                         $text = $commentState->replace( $text );
 667                 }
 668
 669                 wfProfileOut( __METHOD__ );
 670                 return $text;
 671         }
 672
 673         /**
 674          * Restores pre, math, and other extensions removed by strip()
 675          *
 676          * always call unstripNoWiki() after this one
 677          * @private
 678          * @deprecated use $this->mStripState->unstrip()
 679          */
 680         function unstrip( $text, $state ) {
 681                 return $state->unstripGeneral( $text );
 682         }
 683
 684         /**
 685          * Always call this after unstrip() to preserve the order
 686          *
 687          * @private
 688          * @deprecated use $this->mStripState->unstrip()
 689          */
 690         function unstripNoWiki( $text, $state ) {
 691                 return $state->unstripNoWiki( $text );
 692         }
 693
 694         /**
 695          * @deprecated use $this->mStripState->unstripBoth()
 696          */
 697         function unstripForHTML( $text ) {
 698                 return $this->mStripState->unstripBoth( $text );
 699         }
 700
 701         /**
 702          * Add an item to the strip state
 703          * Returns the unique tag which must be inserted into the stripped text
 704          * The tag will be replaced with the original text in unstrip()
 705          *
 706          * @private
 707          */
 708         function insertStripItem( $text, &$state ) {
 709                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 710                 $state->general->setPair( $rnd, $text );
 711                 return $rnd;
 712         }
 713
 714         /**
 715          * Interface with html tidy, used if $wgUseTidy = true.
 716          * If tidy isn't able to correct the markup, the original will be
 717          * returned in all its glory with a warning comment appended.
 718          *
 719          * Either the external tidy program or the in-process tidy extension
 720          * will be used depending on availability. Override the default
 721          * $wgTidyInternal setting to disable the internal if it's not working.
 722          *
 723          * @param string $text Hideous HTML input
 724          * @return string Corrected HTML output
 725          * @public
 726          * @static
 727          */
 728         function tidy( $text ) {
 729                 global $wgTidyInternal;
 730                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 731 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 732 '<head><title>test</title></head><body>'.$text.'</body></html>';
 733                 if( $wgTidyInternal ) {
 734                         $correctedtext = Parser::internalTidy( $wrappedtext );
 735                 } else {
 736                         $correctedtext = Parser::externalTidy( $wrappedtext );
 737                 }
 738                 if( is_null( $correctedtext ) ) {
 739                         wfDebug( "Tidy error detected!\n" );
 740                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 741                 }
 742                 return $correctedtext;
 743         }
 744
 745         /**
 746          * Spawn an external HTML tidy process and get corrected markup back from it.
 747          *
 748          * @private
 749          * @static
 750          */
 751         function externalTidy( $text ) {
 752                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 753                 $fname = 'Parser::externalTidy';
 754                 wfProfileIn( $fname );
 755
 756                 $cleansource = '';
 757                 $opts = ' -utf8';
 758
 759                 $descriptorspec = array(
 760                         0 => array('pipe', 'r'),
 761                         1 => array('pipe', 'w'),
 762                         2 => array('file', '/dev/null', 'a')  // FIXME: this line in UNIX-specific, it generates a warning on Windows, because /dev/null is not a valid Windows file.
 763                 );
 764                 $pipes = array();
 765                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 766                 if (is_resource($process)) {
 767                         // Theoretically, this style of communication could cause a deadlock
 768                         // here. If the stdout buffer fills up, then writes to stdin could
 769                         // block. This doesn't appear to happen with tidy, because tidy only
 770                         // writes to stdout after it's finished reading from stdin. Search
 771                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 772                         fwrite($pipes[0], $text);
 773                         fclose($pipes[0]);
 774                         while (!feof($pipes[1])) {
 775                                 $cleansource .= fgets($pipes[1], 1024);
 776                         }
 777                         fclose($pipes[1]);
 778                         proc_close($process);
 779                 }
 780
 781                 wfProfileOut( $fname );
 782
 783                 if( $cleansource == '' && $text != '') {
 784                         // Some kind of error happened, so we couldn't get the corrected text.
 785                         // Just give up; we'll use the source text and append a warning.
 786                         return null;
 787                 } else {
 788                         return $cleansource;
 789                 }
 790         }
 791
 792         /**
 793          * Use the HTML tidy PECL extension to use the tidy library in-process,
 794          * saving the overhead of spawning a new process. Currently written to
 795          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 796          *
 797          * 'pear install tidy' should be able to compile the extension module.
 798          *
 799          * @private
 800          * @static
 801          */
 802         function internalTidy( $text ) {
 803                 global $wgTidyConf;
 804                 $fname = 'Parser::internalTidy';
 805                 wfProfileIn( $fname );
 806
 807                 tidy_load_config( $wgTidyConf );
 808                 tidy_set_encoding( 'utf8' );
 809                 tidy_parse_string( $text );
 810                 tidy_clean_repair();
 811                 if( tidy_get_status() == 2 ) {
 812                         // 2 is magic number for fatal error
 813                         // http://www.php.net/manual/en/function.tidy-get-status.php
 814                         $cleansource = null;
 815                 } else {
 816                         $cleansource = tidy_get_output();
 817                 }
 818                 wfProfileOut( $fname );
 819                 return $cleansource;
 820         }
 821
 822         /**
 823          * parse the wiki syntax used to render tables
 824          *
 825          * @private
 826          */
 827         function doTableStuff ( $text ) {
 828                 $fname = 'Parser::doTableStuff';
 829                 wfProfileIn( $fname );
 830
 831                 $lines = explode ( "\n" , $text );
 832                 $td_history = array (); // Is currently a td tag open?
 833                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 834                 $tr_history = array (); // Is currently a tr tag open?
 835                 $tr_attributes = array (); // history of tr attributes
 836                 $has_opened_tr = array(); // Did this table open a <tr> element?
 837                 $indent_level = 0; // indent level of the table
 838                 foreach ( $lines as $key => $line )
 839                 {
 840                         $line = trim ( $line );
 841
 842                         if( $line == '' ) { // empty line, go to next line
 843                                 continue;
 844                         }
 845                         $first_character = $line{0};
 846                         $matches = array();
 847
 848                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 849                                 // First check if we are starting a new table
 850                                 $indent_level = strlen( $matches[1] );
 851
 852                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 853                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 854
 855                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 856                                 array_push ( $td_history , false );
 857                                 array_push ( $last_tag_history , '' );
 858                                 array_push ( $tr_history , false );
 859                                 array_push ( $tr_attributes , '' );
 860                                 array_push ( $has_opened_tr , false );
 861                         } else if ( count ( $td_history ) == 0 ) {
 862                                 // Don't do any of the following
 863                                 continue;
 864                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 865                                 // We are ending a table
 866                                 $line = '</table>' . substr ( $line , 2 );
 867                                 $last_tag = array_pop ( $last_tag_history );
 868
 869                                 if ( !array_pop ( $has_opened_tr ) ) {
 870                                         $line = "<tr><td></td></tr>{$line}";
 871                                 }
 872
 873                                 if ( array_pop ( $tr_history ) ) {
 874                                         $line = "</tr>{$line}";
 875                                 }
 876
 877                                 if ( array_pop ( $td_history ) ) {
 878                                         $line = "</{$last_tag}>{$line}";
 879                                 }
 880                                 array_pop ( $tr_attributes );
 881                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 882                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 883                                 // Now we have a table row
 884                                 $line = preg_replace( '#^\|-+#', '', $line );
 885
 886                                 // Whats after the tag is now only attributes
 887                                 $attributes = $this->mStripState->unstripBoth( $line );
 888                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 889                                 array_pop ( $tr_attributes );
 890                                 array_push ( $tr_attributes , $attributes );
 891
 892                                 $line = '';
 893                                 $last_tag = array_pop ( $last_tag_history );
 894                                 array_pop ( $has_opened_tr );
 895                                 array_push ( $has_opened_tr , true );
 896
 897                                 if ( array_pop ( $tr_history ) ) {
 898                                         $line = '</tr>';
 899                                 }
 900
 901                                 if ( array_pop ( $td_history ) ) {
 902                                         $line = "</{$last_tag}>{$line}";
 903                                 }
 904
 905                                 $lines[$key] = $line;
 906                                 array_push ( $tr_history , false );
 907                                 array_push ( $td_history , false );
 908                                 array_push ( $last_tag_history , '' );
 909                         }
 910                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 911                                 // This might be cell elements, td, th or captions
 912                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 913                                         $first_character = '+';
 914                                         $line = substr ( $line , 1 );
 915                                 }
 916
 917                                 $line = substr ( $line , 1 );
 918
 919                                 if ( $first_character == '!' ) {
 920                                         $line = str_replace ( '!!' , '||' , $line );
 921                                 }
 922
 923                                 // Split up multiple cells on the same line.
 924                                 // FIXME : This can result in improper nesting of tags processed
 925                                 // by earlier parser steps, but should avoid splitting up eg
 926                                 // attribute values containing literal "||".
 927                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 928
 929                                 $lines[$key] = '';
 930
 931                                 // Loop through each table cell
 932                                 foreach ( $cells as $cell )
 933                                 {
 934                                         $previous = '';
 935                                         if ( $first_character != '+' )
 936                                         {
 937                                                 $tr_after = array_pop ( $tr_attributes );
 938                                                 if ( !array_pop ( $tr_history ) ) {
 939                                                         $previous = "<tr{$tr_after}>\n";
 940                                                 }
 941                                                 array_push ( $tr_history , true );
 942                                                 array_push ( $tr_attributes , '' );
 943                                                 array_pop ( $has_opened_tr );
 944                                                 array_push ( $has_opened_tr , true );
 945                                         }
 946
 947                                         $last_tag = array_pop ( $last_tag_history );
 948
 949                                         if ( array_pop ( $td_history ) ) {
 950                                                 $previous = "</{$last_tag}>{$previous}";
 951                                         }
 952
 953                                         if ( $first_character == '|' ) {
 954                                                 $last_tag = 'td';
 955                                         } else if ( $first_character == '!' ) {
 956                                                 $last_tag = 'th';
 957                                         } else if ( $first_character == '+' ) {
 958                                                 $last_tag = 'caption';
 959                                         } else {
 960                                                 $last_tag = '';
 961                                         }
 962
 963                                         array_push ( $last_tag_history , $last_tag );
 964
 965                                         // A cell could contain both parameters and data
 966                                         $cell_data = explode ( '|' , $cell , 2 );
 967
 968                                         // Bug 553: Note that a '|' inside an invalid link should not
 969                                         // be mistaken as delimiting cell parameters
 970                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 971                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 972                                         } else if ( count ( $cell_data ) == 1 )
 973                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 974                                         else {
 975                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 976                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 977                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 978                                         }
 979
 980                                         $lines[$key] .= $cell;
 981                                         array_push ( $td_history , true );
 982                                 }
 983                         }
 984                 }
 985
 986                 // Closing open td, tr && table
 987                 while ( count ( $td_history ) > 0 )
 988                 {
 989                         if ( array_pop ( $td_history ) ) {
 990                                 $lines[] = '</td>' ;
 991                         }
 992                         if ( array_pop ( $tr_history ) ) {
 993                                 $lines[] = '</tr>' ;
 994                         }
 995                         if ( !array_pop ( $has_opened_tr ) ) {
 996                                 $lines[] = "<tr><td></td></tr>" ;
 997                         }
 998
 999                         $lines[] = '</table>' ;
1000                 }
1001
1002                 $output = implode ( "\n" , $lines ) ;
1003
1004                 // special case: don't return empty table
1005                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
1006                         $output = '';
1007                 }
1008
1009                 wfProfileOut( $fname );
1010
1011                 return $output;
1012         }
1013
1014         /**
1015          * Helper function for parse() that transforms wiki markup into
1016          * HTML. Only called for $mOutputType == OT_HTML.
1017          *
1018          * @private
1019          */
1020         function internalParse( $text ) {
1021                 $args = array();
1022                 $isMain = true;
1023                 $fname = 'Parser::internalParse';
1024                 wfProfileIn( $fname );
1025
1026                 # Hook to suspend the parser in this state
1027                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1028                         wfProfileOut( $fname );
1029                         return $text ;
1030                 }
1031
1032                 # Remove <noinclude> tags and <includeonly> sections
1033                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
1034                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
1035                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
1036
1037                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
1038
1039                 $text = $this->replaceVariables( $text, $args );
1040                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1041
1042                 // Tables need to come after variable replacement for things to work
1043                 // properly; putting them before other transformations should keep
1044                 // exciting things like link expansions from showing up in surprising
1045                 // places.
1046                 $text = $this->doTableStuff( $text );
1047
1048                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1049
1050                 $text = $this->stripToc( $text );
1051                 $this->stripNoGallery( $text );
1052                 $text = $this->doHeadings( $text );
1053                 if($this->mOptions->getUseDynamicDates()) {
1054                         $df =& DateFormatter::getInstance();
1055                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1056                 }
1057                 $text = $this->doAllQuotes( $text );
1058                 $text = $this->replaceInternalLinks( $text );
1059                 $text = $this->replaceExternalLinks( $text );
1060
1061                 # replaceInternalLinks may sometimes leave behind
1062                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1063                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1064
1065                 $text = $this->doMagicLinks( $text );
1066                 $text = $this->formatHeadings( $text, $isMain );
1067
1068                 wfProfileOut( $fname );
1069                 return $text;
1070         }
1071
1072         /**
1073          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1074          * magic external links.
1075          *
1076          * @private
1077          */
1078         function &doMagicLinks( &$text ) {
1079                 wfProfileIn( __METHOD__ );
1080                 $text = preg_replace_callback(
1081                         '!(?:                           # Start cases
1082                             <a.*?</a> |                 # Skip link text
1083                             <.*?> |                     # Skip stuff inside HTML elements
1084                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1085                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1086                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1087                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1088                                      [0-9Xx]                 # check digit
1089                                    \b)
1090                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1091                 wfProfileOut( __METHOD__ );
1092                 return $text;
1093         }
1094
1095         function magicLinkCallback( $m ) {
1096                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1097                         # Skip HTML element
1098                         return $m[0];
1099                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1100                         $isbn = $m[2];
1101                         $num = strtr( $isbn, array(
1102                                 '-' => '',
1103                                 ' ' => '',
1104                                 'x' => 'X',
1105                         ));
1106                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1107                         $text = '<a href="' .
1108                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1109                                 "\" class=\"internal\">ISBN $isbn</a>";
1110                 } else {
1111                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1112                                 $keyword = 'RFC';
1113                                 $urlmsg = 'rfcurl';
1114                                 $id = $m[1];
1115                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1116                                 $keyword = 'PMID';
1117                                 $urlmsg = 'pubmedurl';
1118                                 $id = $m[1];
1119                         } else {
1120                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1121                                         substr($m[0], 0, 20 ) . '"' );
1122                         }
1123
1124                         $url = wfMsg( $urlmsg, $id);
1125                         $sk = $this->mOptions->getSkin();
1126                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1127                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1128                 }
1129                 return $text;
1130         }
1131
1132         /**
1133          * Parse headers and return html
1134          *
1135          * @private
1136          */
1137         function doHeadings( $text ) {
1138                 $fname = 'Parser::doHeadings';
1139                 wfProfileIn( $fname );
1140                 for ( $i = 6; $i >= 1; --$i ) {
1141                         $h = str_repeat( '=', $i );
1142                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1143                           "<h{$i}>\\1</h{$i}>\\2", $text );
1144                 }
1145                 wfProfileOut( $fname );
1146                 return $text;
1147         }
1148
1149         /**
1150          * Replace single quotes with HTML markup
1151          * @private
1152          * @return string the altered text
1153          */
1154         function doAllQuotes( $text ) {
1155                 $fname = 'Parser::doAllQuotes';
1156                 wfProfileIn( $fname );
1157                 $outtext = '';
1158                 $lines = explode( "\n", $text );
1159                 foreach ( $lines as $line ) {
1160                         $outtext .= $this->doQuotes ( $line ) . "\n";
1161                 }
1162                 $outtext = substr($outtext, 0,-1);
1163                 wfProfileOut( $fname );
1164                 return $outtext;
1165         }
1166
1167         /**
1168          * Helper function for doAllQuotes()
1169          * @private
1170          */
1171         function doQuotes( $text ) {
1172                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1173                 if ( count( $arr ) == 1 )
1174                         return $text;
1175                 else
1176                 {
1177                         # First, do some preliminary work. This may shift some apostrophes from
1178                         # being mark-up to being text. It also counts the number of occurrences
1179                         # of bold and italics mark-ups.
1180                         $i = 0;
1181                         $numbold = 0;
1182                         $numitalics = 0;
1183                         foreach ( $arr as $r )
1184                         {
1185                                 if ( ( $i % 2 ) == 1 )
1186                                 {
1187                                         # If there are ever four apostrophes, assume the first is supposed to
1188                                         # be text, and the remaining three constitute mark-up for bold text.
1189                                         if ( strlen( $arr[$i] ) == 4 )
1190                                         {
1191                                                 $arr[$i-1] .= "'";
1192                                                 $arr[$i] = "'''";
1193                                         }
1194                                         # If there are more than 5 apostrophes in a row, assume they're all
1195                                         # text except for the last 5.
1196                                         else if ( strlen( $arr[$i] ) > 5 )
1197                                         {
1198                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1199                                                 $arr[$i] = "'''''";
1200                                         }
1201                                         # Count the number of occurrences of bold and italics mark-ups.
1202                                         # We are not counting sequences of five apostrophes.
1203                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1204                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1205                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1206                                 }
1207                                 $i++;
1208                         }
1209
1210                         # If there is an odd number of both bold and italics, it is likely
1211                         # that one of the bold ones was meant to be an apostrophe followed
1212                         # by italics. Which one we cannot know for certain, but it is more
1213                         # likely to be one that has a single-letter word before it.
1214                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1215                         {
1216                                 $i = 0;
1217                                 $firstsingleletterword = -1;
1218                                 $firstmultiletterword = -1;
1219                                 $firstspace = -1;
1220                                 foreach ( $arr as $r )
1221                                 {
1222                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1223                                         {
1224                                                 $x1 = substr ($arr[$i-1], -1);
1225                                                 $x2 = substr ($arr[$i-1], -2, 1);
1226                                                 if ($x1 == ' ') {
1227                                                         if ($firstspace == -1) $firstspace = $i;
1228                                                 } else if ($x2 == ' ') {
1229                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1230                                                 } else {
1231                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1232                                                 }
1233                                         }
1234                                         $i++;
1235                                 }
1236
1237                                 # If there is a single-letter word, use it!
1238                                 if ($firstsingleletterword > -1)
1239                                 {
1240                                         $arr [ $firstsingleletterword ] = "''";
1241                                         $arr [ $firstsingleletterword-1 ] .= "'";
1242                                 }
1243                                 # If not, but there's a multi-letter word, use that one.
1244                                 else if ($firstmultiletterword > -1)
1245                                 {
1246                                         $arr [ $firstmultiletterword ] = "''";
1247                                         $arr [ $firstmultiletterword-1 ] .= "'";
1248                                 }
1249                                 # ... otherwise use the first one that has neither.
1250                                 # (notice that it is possible for all three to be -1 if, for example,
1251                                 # there is only one pentuple-apostrophe in the line)
1252                                 else if ($firstspace > -1)
1253                                 {
1254                                         $arr [ $firstspace ] = "''";
1255                                         $arr [ $firstspace-1 ] .= "'";
1256                                 }
1257                         }
1258
1259                         # Now let's actually convert our apostrophic mush to HTML!
1260                         $output = '';
1261                         $buffer = '';
1262                         $state = '';
1263                         $i = 0;
1264                         foreach ($arr as $r)
1265                         {
1266                                 if (($i % 2) == 0)
1267                                 {
1268                                         if ($state == 'both')
1269                                                 $buffer .= $r;
1270                                         else
1271                                                 $output .= $r;
1272                                 }
1273                                 else
1274                                 {
1275                                         if (strlen ($r) == 2)
1276                                         {
1277                                                 if ($state == 'i')
1278                                                 { $output .= '</i>'; $state = ''; }
1279                                                 else if ($state == 'bi')
1280                                                 { $output .= '</i>'; $state = 'b'; }
1281                                                 else if ($state == 'ib')
1282                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1283                                                 else if ($state == 'both')
1284                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1285                                                 else # $state can be 'b' or ''
1286                                                 { $output .= '<i>'; $state .= 'i'; }
1287                                         }
1288                                         else if (strlen ($r) == 3)
1289                                         {
1290                                                 if ($state == 'b')
1291                                                 { $output .= '</b>'; $state = ''; }
1292                                                 else if ($state == 'bi')
1293                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1294                                                 else if ($state == 'ib')
1295                                                 { $output .= '</b>'; $state = 'i'; }
1296                                                 else if ($state == 'both')
1297                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1298                                                 else # $state can be 'i' or ''
1299                                                 { $output .= '<b>'; $state .= 'b'; }
1300                                         }
1301                                         else if (strlen ($r) == 5)
1302                                         {
1303                                                 if ($state == 'b')
1304                                                 { $output .= '</b><i>'; $state = 'i'; }
1305                                                 else if ($state == 'i')
1306                                                 { $output .= '</i><b>'; $state = 'b'; }
1307                                                 else if ($state == 'bi')
1308                                                 { $output .= '</i></b>'; $state = ''; }
1309                                                 else if ($state == 'ib')
1310                                                 { $output .= '</b></i>'; $state = ''; }
1311                                                 else if ($state == 'both')
1312                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1313                                                 else # ($state == '')
1314                                                 { $buffer = ''; $state = 'both'; }
1315                                         }
1316                                 }
1317                                 $i++;
1318                         }
1319                         # Now close all remaining tags.  Notice that the order is important.
1320                         if ($state == 'b' || $state == 'ib')
1321                                 $output .= '</b>';
1322                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1323                                 $output .= '</i>';
1324                         if ($state == 'bi')
1325                                 $output .= '</b>';
1326                         # There might be lonely ''''', so make sure we have a buffer
1327                         if ($state == 'both' && $buffer)
1328                                 $output .= '<b><i>'.$buffer.'</i></b>';
1329                         return $output;
1330                 }
1331         }
1332
1333         /**
1334          * Replace external links
1335          *
1336          * Note: this is all very hackish and the order of execution matters a lot.
1337          * Make sure to run maintenance/parserTests.php if you change this code.
1338          *
1339          * @private
1340          */
1341         function replaceExternalLinks( $text ) {
1342                 global $wgContLang;
1343                 $fname = 'Parser::replaceExternalLinks';
1344                 wfProfileIn( $fname );
1345
1346                 $sk = $this->mOptions->getSkin();
1347
1348                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1349
1350                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1351
1352                 $i = 0;
1353                 while ( $i<count( $bits ) ) {
1354                         $url = $bits[$i++];
1355                         $protocol = $bits[$i++];
1356                         $text = $bits[$i++];
1357                         $trail = $bits[$i++];
1358
1359                         # The characters '<' and '>' (which were escaped by
1360                         # removeHTMLtags()) should not be included in
1361                         # URLs, per RFC 2396.
1362                         $m2 = array();
1363                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1364                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1365                                 $url = substr($url, 0, $m2[0][1]);
1366                         }
1367
1368                         # If the link text is an image URL, replace it with an <img> tag
1369                         # This happened by accident in the original parser, but some people used it extensively
1370                         $img = $this->maybeMakeExternalImage( $text );
1371                         if ( $img !== false ) {
1372                                 $text = $img;
1373                         }
1374
1375                         $dtrail = '';
1376
1377                         # Set linktype for CSS - if URL==text, link is essentially free
1378                         $linktype = ($text == $url) ? 'free' : 'text';
1379
1380                         # No link text, e.g. [http://domain.tld/some.link]
1381                         if ( $text == '' ) {
1382                                 # Autonumber if allowed. See bug #5918
1383                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1384                                         $text = '[' . ++$this->mAutonumber . ']';
1385                                         $linktype = 'autonumber';
1386                                 } else {
1387                                         # Otherwise just use the URL
1388                                         $text = htmlspecialchars( $url );
1389                                         $linktype = 'free';
1390                                 }
1391                         } else {
1392                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1393                                 # Check for trail
1394                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1395                         }
1396
1397                         $text = $wgContLang->markNoConversion($text);
1398
1399                         $url = Sanitizer::cleanUrl( $url );
1400
1401                         # Process the trail (i.e. everything after this link up until start of the next link),
1402                         # replacing any non-bracketed links
1403                         $trail = $this->replaceFreeExternalLinks( $trail );
1404
1405                         # Use the encoded URL
1406                         # This means that users can paste URLs directly into the text
1407                         # Funny characters like &ouml; aren't valid in URLs anyway
1408                         # This was changed in August 2004
1409                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1410
1411                         # Register link in the output object.
1412                         # Replace unnecessary URL escape codes with the referenced character
1413                         # This prevents spammers from hiding links from the filters
1414                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1415                         $this->mOutput->addExternalLink( $pasteurized );
1416                 }
1417
1418                 wfProfileOut( $fname );
1419                 return $s;
1420         }
1421
1422         /**
1423          * Replace anything that looks like a URL with a link
1424          * @private
1425          */
1426         function replaceFreeExternalLinks( $text ) {
1427                 global $wgContLang;
1428                 $fname = 'Parser::replaceFreeExternalLinks';
1429                 wfProfileIn( $fname );
1430
1431                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1432                 $s = array_shift( $bits );
1433                 $i = 0;
1434
1435                 $sk = $this->mOptions->getSkin();
1436
1437                 while ( $i < count( $bits ) ){
1438                         $protocol = $bits[$i++];
1439                         $remainder = $bits[$i++];
1440
1441                         $m = array();
1442                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1443                                 # Found some characters after the protocol that look promising
1444                                 $url = $protocol . $m[1];
1445                                 $trail = $m[2];
1446
1447                                 # special case: handle urls as url args:
1448                                 # http://www.example.com/foo?=http://www.example.com/bar
1449                                 if(strlen($trail) == 0 &&
1450                                         isset($bits[$i]) &&
1451                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1452                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1453                                 {
1454                                         # add protocol, arg
1455                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1456                                         $i += 2;
1457                                         $trail = $m[2];
1458                                 }
1459
1460                                 # The characters '<' and '>' (which were escaped by
1461                                 # removeHTMLtags()) should not be included in
1462                                 # URLs, per RFC 2396.
1463                                 $m2 = array();
1464                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1465                                         $trail = substr($url, $m2[0][1]) . $trail;
1466                                         $url = substr($url, 0, $m2[0][1]);
1467                                 }
1468
1469                                 # Move trailing punctuation to $trail
1470                                 $sep = ',;\.:!?';
1471                                 # If there is no left bracket, then consider right brackets fair game too
1472                                 if ( strpos( $url, '(' ) === false ) {
1473                                         $sep .= ')';
1474                                 }
1475
1476                                 $numSepChars = strspn( strrev( $url ), $sep );
1477                                 if ( $numSepChars ) {
1478                                         $trail = substr( $url, -$numSepChars ) . $trail;
1479                                         $url = substr( $url, 0, -$numSepChars );
1480                                 }
1481
1482                                 $url = Sanitizer::cleanUrl( $url );
1483
1484                                 # Is this an external image?
1485                                 $text = $this->maybeMakeExternalImage( $url );
1486                                 if ( $text === false ) {
1487                                         # Not an image, make a link
1488                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1489                                         # Register it in the output object...
1490                                         # Replace unnecessary URL escape codes with their equivalent characters
1491                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1492                                         $this->mOutput->addExternalLink( $pasteurized );
1493                                 }
1494                                 $s .= $text . $trail;
1495                         } else {
1496                                 $s .= $protocol . $remainder;
1497                         }
1498                 }
1499                 wfProfileOut( $fname );
1500                 return $s;
1501         }
1502
1503         /**
1504          * Replace unusual URL escape codes with their equivalent characters
1505          * @param string
1506          * @return string
1507          * @static
1508          * @todo  This can merge genuinely required bits in the path or query string,
1509          *        breaking legit URLs. A proper fix would treat the various parts of
1510          *        the URL differently; as a workaround, just use the output for
1511          *        statistical records, not for actual linking/output.
1512          */
1513         static function replaceUnusualEscapes( $url ) {
1514                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1515                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1516         }
1517
1518         /**
1519          * Callback function used in replaceUnusualEscapes().
1520          * Replaces unusual URL escape codes with their equivalent character
1521          * @static
1522          * @private
1523          */
1524         private static function replaceUnusualEscapesCallback( $matches ) {
1525                 $char = urldecode( $matches[0] );
1526                 $ord = ord( $char );
1527                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1528                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1529                         // No, shouldn't be escaped
1530                         return $char;
1531                 } else {
1532                         // Yes, leave it escaped
1533                         return $matches[0];
1534                 }
1535         }
1536
1537         /**
1538          * make an image if it's allowed, either through the global
1539          * option or through the exception
1540          * @private
1541          */
1542         function maybeMakeExternalImage( $url ) {
1543                 $sk = $this->mOptions->getSkin();
1544                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1545                 $imagesexception = !empty($imagesfrom);
1546                 $text = false;
1547                 if ( $this->mOptions->getAllowExternalImages()
1548                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1549                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1550                                 # Image found
1551                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1552                         }
1553                 }
1554                 return $text;
1555         }
1556
1557         /**
1558          * Process [[ ]] wikilinks
1559          *
1560          * @private
1561          */
1562         function replaceInternalLinks( $s ) {
1563                 global $wgContLang;
1564                 static $fname = 'Parser::replaceInternalLinks' ;
1565
1566                 wfProfileIn( $fname );
1567
1568                 wfProfileIn( $fname.'-setup' );
1569                 static $tc = FALSE;
1570                 # the % is needed to support urlencoded titles as well
1571                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1572
1573                 $sk = $this->mOptions->getSkin();
1574
1575                 #split the entire text string on occurences of [[
1576                 $a = explode( '[[', ' ' . $s );
1577                 #get the first element (all text up to first [[), and remove the space we added
1578                 $s = array_shift( $a );
1579                 $s = substr( $s, 1 );
1580
1581                 # Match a link having the form [[namespace:link|alternate]]trail
1582                 static $e1 = FALSE;
1583                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1584                 # Match cases where there is no "]]", which might still be images
1585                 static $e1_img = FALSE;
1586                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1587                 # Match the end of a line for a word that's not followed by whitespace,
1588                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1589                 $e2 = wfMsgForContent( 'linkprefix' );
1590
1591                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1592                 if( is_null( $this->mTitle ) ) {
1593                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1594                 }
1595                 $nottalk = !$this->mTitle->isTalkPage();
1596
1597                 if ( $useLinkPrefixExtension ) {
1598                         $m = array();
1599                         if ( preg_match( $e2, $s, $m ) ) {
1600                                 $first_prefix = $m[2];
1601                         } else {
1602                                 $first_prefix = false;
1603                         }
1604                 } else {
1605                         $prefix = '';
1606                 }
1607
1608                 if($wgContLang->hasVariants()) {
1609                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1610                 } else {
1611                         $selflink = array($this->mTitle->getPrefixedText());
1612                 }
1613                 $useSubpages = $this->areSubpagesAllowed();
1614                 wfProfileOut( $fname.'-setup' );
1615
1616                 # Loop for each link
1617                 for ($k = 0; isset( $a[$k] ); $k++) {
1618                         $line = $a[$k];
1619                         if ( $useLinkPrefixExtension ) {
1620                                 wfProfileIn( $fname.'-prefixhandling' );
1621                                 if ( preg_match( $e2, $s, $m ) ) {
1622                                         $prefix = $m[2];
1623                                         $s = $m[1];
1624                                 } else {
1625                                         $prefix='';
1626                                 }
1627                                 # first link
1628                                 if($first_prefix) {
1629                                         $prefix = $first_prefix;
1630                                         $first_prefix = false;
1631                                 }
1632                                 wfProfileOut( $fname.'-prefixhandling' );
1633                         }
1634
1635                         $might_be_img = false;
1636
1637                         wfProfileIn( "$fname-e1" );
1638                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1639                                 $text = $m[2];
1640                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1641                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1642                                 # the real problem is with the $e1 regex
1643                                 # See bug 1300.
1644                                 #
1645                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1646                                 # and no image is in sight. See bug 2095.
1647                                 #
1648                                 if( $text !== '' &&
1649                                         substr( $m[3], 0, 1 ) === ']' &&
1650                                         strpos($text, '[') !== false
1651                                 )
1652                                 {
1653                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1654                                         $m[3] = substr( $m[3], 1 );
1655                                 }
1656                                 # fix up urlencoded title texts
1657                                 if( strpos( $m[1], '%' ) !== false ) {
1658                                         # Should anchors '#' also be rejected?
1659                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1660                                 }
1661                                 $trail = $m[3];
1662                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1663                                 $might_be_img = true;
1664                                 $text = $m[2];
1665                                 if ( strpos( $m[1], '%' ) !== false ) {
1666                                         $m[1] = urldecode($m[1]);
1667                                 }
1668                                 $trail = "";
1669                         } else { # Invalid form; output directly
1670                                 $s .= $prefix . '[[' . $line ;
1671                                 wfProfileOut( "$fname-e1" );
1672                                 continue;
1673                         }
1674                         wfProfileOut( "$fname-e1" );
1675                         wfProfileIn( "$fname-misc" );
1676
1677                         # Don't allow internal links to pages containing
1678                         # PROTO: where PROTO is a valid URL protocol; these
1679                         # should be external links.
1680                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1681                                 $s .= $prefix . '[[' . $line ;
1682                                 continue;
1683                         }
1684
1685                         # Make subpage if necessary
1686                         if( $useSubpages ) {
1687                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1688                         } else {
1689                                 $link = $m[1];
1690                         }
1691
1692                         $noforce = (substr($m[1], 0, 1) != ':');
1693                         if (!$noforce) {
1694                                 # Strip off leading ':'
1695                                 $link = substr($link, 1);
1696                         }
1697
1698                         wfProfileOut( "$fname-misc" );
1699                         wfProfileIn( "$fname-title" );
1700                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1701                         if( !$nt ) {
1702                                 $s .= $prefix . '[[' . $line;
1703                                 wfProfileOut( "$fname-title" );
1704                                 continue;
1705                         }
1706
1707                         $ns = $nt->getNamespace();
1708                         $iw = $nt->getInterWiki();
1709                         wfProfileOut( "$fname-title" );
1710
1711                         if ($might_be_img) { # if this is actually an invalid link
1712                                 wfProfileIn( "$fname-might_be_img" );
1713                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1714                                         $found = false;
1715                                         while (isset ($a[$k+1]) ) {
1716                                                 #look at the next 'line' to see if we can close it there
1717                                                 $spliced = array_splice( $a, $k + 1, 1 );
1718                                                 $next_line = array_shift( $spliced );
1719                                                 $m = explode( ']]', $next_line, 3 );
1720                                                 if ( count( $m ) == 3 ) {
1721                                                         # the first ]] closes the inner link, the second the image
1722                                                         $found = true;
1723                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1724                                                         $trail = $m[2];
1725                                                         break;
1726                                                 } elseif ( count( $m ) == 2 ) {
1727                                                         #if there's exactly one ]] that's fine, we'll keep looking
1728                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1729                                                 } else {
1730                                                         #if $next_line is invalid too, we need look no further
1731                                                         $text .= '[[' . $next_line;
1732                                                         break;
1733                                                 }
1734                                         }
1735                                         if ( !$found ) {
1736                                                 # we couldn't find the end of this imageLink, so output it raw
1737                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1738                                                 $text = $this->replaceInternalLinks($text);
1739                                                 $s .= "{$prefix}[[$link|$text";
1740                                                 # note: no $trail, because without an end, there *is* no trail
1741                                                 wfProfileOut( "$fname-might_be_img" );
1742                                                 continue;
1743                                         }
1744                                 } else { #it's not an image, so output it raw
1745                                         $s .= "{$prefix}[[$link|$text";
1746                                         # note: no $trail, because without an end, there *is* no trail
1747                                         wfProfileOut( "$fname-might_be_img" );
1748                                         continue;
1749                                 }
1750                                 wfProfileOut( "$fname-might_be_img" );
1751                         }
1752
1753                         $wasblank = ( '' == $text );
1754                         if( $wasblank ) $text = $link;
1755
1756                         # Link not escaped by : , create the various objects
1757                         if( $noforce ) {
1758
1759                                 # Interwikis
1760                                 wfProfileIn( "$fname-interwiki" );
1761                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1762                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1763                                         $s = rtrim($s . $prefix);
1764                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1765                                         wfProfileOut( "$fname-interwiki" );
1766                                         continue;
1767                                 }
1768                                 wfProfileOut( "$fname-interwiki" );
1769
1770                                 if ( $ns == NS_IMAGE ) {
1771                                         wfProfileIn( "$fname-image" );
1772                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1773                                                 # recursively parse links inside the image caption
1774                                                 # actually, this will parse them in any other parameters, too,
1775                                                 # but it might be hard to fix that, and it doesn't matter ATM
1776                                                 $text = $this->replaceExternalLinks($text);
1777                                                 $text = $this->replaceInternalLinks($text);
1778
1779                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1780                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1781                                                 $this->mOutput->addImage( $nt->getDBkey() );
1782
1783                                                 wfProfileOut( "$fname-image" );
1784                                                 continue;
1785                                         } else {
1786                                                 # We still need to record the image's presence on the page
1787                                                 $this->mOutput->addImage( $nt->getDBkey() );
1788                                         }
1789                                         wfProfileOut( "$fname-image" );
1790
1791                                 }
1792
1793                                 if ( $ns == NS_CATEGORY ) {
1794                                         wfProfileIn( "$fname-category" );
1795                                         $s = rtrim($s . "\n"); # bug 87
1796
1797                                         if ( $wasblank ) {
1798                                                 $sortkey = $this->getDefaultSort();
1799                                         } else {
1800                                                 $sortkey = $text;
1801                                         }
1802                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1803                                         $sortkey = str_replace( "\n", '', $sortkey );
1804                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1805                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1806
1807                                         /**
1808                                          * Strip the whitespace Category links produce, see bug 87
1809                                          * @todo We might want to use trim($tmp, "\n") here.
1810                                          */
1811                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1812
1813                                         wfProfileOut( "$fname-category" );
1814                                         continue;
1815                                 }
1816                         }
1817
1818                         # Self-link checking
1819                         if( $nt->getFragment() === '' ) {
1820                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1821                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1822                                         continue;
1823                                 }
1824                         }
1825
1826                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1827                         if( $ns == NS_MEDIA ) {
1828                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1829                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1830                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1831                                 $this->mOutput->addImage( $nt->getDBkey() );
1832                                 continue;
1833                         } elseif( $ns == NS_SPECIAL ) {
1834                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1835                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1836                                 } else {
1837                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1838                                 }
1839                                 continue;
1840                         } elseif( $ns == NS_IMAGE ) {
1841                                 $img = wfFindFile( $nt );
1842                                 if( $img ) {
1843                                         // Force a blue link if the file exists; may be a remote
1844                                         // upload on the shared repository, and we want to see its
1845                                         // auto-generated page.
1846                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1847                                         $this->mOutput->addLink( $nt );
1848                                         continue;
1849                                 }
1850                         }
1851                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1852                 }
1853                 wfProfileOut( $fname );
1854                 return $s;
1855         }
1856
1857         /**
1858          * Make a link placeholder. The text returned can be later resolved to a real link with
1859          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1860          * parsing of interwiki links, and secondly to allow all existence checks and
1861          * article length checks (for stub links) to be bundled into a single query.
1862          *
1863          */
1864         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1865                 wfProfileIn( __METHOD__ );
1866                 if ( ! is_object($nt) ) {
1867                         # Fail gracefully
1868                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1869                 } else {
1870                         # Separate the link trail from the rest of the link
1871                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1872
1873                         if ( $nt->isExternal() ) {
1874                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1875                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1876                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1877                         } else {
1878                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1879                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1880                                 $this->mLinkHolders['queries'][] = $query;
1881                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1882                                 $this->mLinkHolders['titles'][] = $nt;
1883
1884                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1885                         }
1886                 }
1887                 wfProfileOut( __METHOD__ );
1888                 return $retVal;
1889         }
1890
1891         /**
1892          * Render a forced-blue link inline; protect against double expansion of
1893          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1894          * Since this little disaster has to split off the trail text to avoid
1895          * breaking URLs in the following text without breaking trails on the
1896          * wiki links, it's been made into a horrible function.
1897          *
1898          * @param Title $nt
1899          * @param string $text
1900          * @param string $query
1901          * @param string $trail
1902          * @param string $prefix
1903          * @return string HTML-wikitext mix oh yuck
1904          */
1905         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1906                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1907                 $sk = $this->mOptions->getSkin();
1908                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1909                 return $this->armorLinks( $link ) . $trail;
1910         }
1911
1912         /**
1913          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1914          * going to go through further parsing steps before inline URL expansion.
1915          *
1916          * In particular this is important when using action=render, which causes
1917          * full URLs to be included.
1918          *
1919          * Oh man I hate our multi-layer parser!
1920          *
1921          * @param string more-or-less HTML
1922          * @return string less-or-more HTML with NOPARSE bits
1923          */
1924         function armorLinks( $text ) {
1925                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1926                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1927         }
1928
1929         /**
1930          * Return true if subpage links should be expanded on this page.
1931          * @return bool
1932          */
1933         function areSubpagesAllowed() {
1934                 # Some namespaces don't allow subpages
1935                 global $wgNamespacesWithSubpages;
1936                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1937         }
1938
1939         /**
1940          * Handle link to subpage if necessary
1941          * @param string $target the source of the link
1942          * @param string &$text the link text, modified as necessary
1943          * @return string the full name of the link
1944          * @private
1945          */
1946         function maybeDoSubpageLink($target, &$text) {
1947                 # Valid link forms:
1948                 # Foobar -- normal
1949                 # :Foobar -- override special treatment of prefix (images, language links)
1950                 # /Foobar -- convert to CurrentPage/Foobar
1951                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1952                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1953                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1954
1955                 $fname = 'Parser::maybeDoSubpageLink';
1956                 wfProfileIn( $fname );
1957                 $ret = $target; # default return value is no change
1958
1959                 # Some namespaces don't allow subpages,
1960                 # so only perform processing if subpages are allowed
1961                 if( $this->areSubpagesAllowed() ) {
1962                         $hash = strpos( $target, '#' );
1963                         if( $hash !== false ) {
1964                                 $suffix = substr( $target, $hash );
1965                                 $target = substr( $target, 0, $hash );
1966                         } else {
1967                                 $suffix = '';
1968                         }
1969                         # bug 7425
1970                         $target = trim( $target );
1971                         # Look at the first character
1972                         if( $target != '' && $target{0} == '/' ) {
1973                                 # / at end means we don't want the slash to be shown
1974                                 $m = array();
1975                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1976                                 if( $trailingSlashes ) {
1977                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1978                                 } else {
1979                                         $noslash = substr( $target, 1 );
1980                                 }
1981
1982                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1983                                 if( '' === $text ) {
1984                                         $text = $target . $suffix;
1985                                 } # this might be changed for ugliness reasons
1986                         } else {
1987                                 # check for .. subpage backlinks
1988                                 $dotdotcount = 0;
1989                                 $nodotdot = $target;
1990                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1991                                         ++$dotdotcount;
1992                                         $nodotdot = substr( $nodotdot, 3 );
1993                                 }
1994                                 if($dotdotcount > 0) {
1995                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1996                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1997                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1998                                                 # / at the end means don't show full path
1999                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
2000                                                         $nodotdot = substr( $nodotdot, 0, -1 );
2001                                                         if( '' === $text ) {
2002                                                                 $text = $nodotdot . $suffix;
2003                                                         }
2004                                                 }
2005                                                 $nodotdot = trim( $nodotdot );
2006                                                 if( $nodotdot != '' ) {
2007                                                         $ret .= '/' . $nodotdot;
2008                                                 }
2009                                                 $ret .= $suffix;
2010                                         }
2011                                 }
2012                         }
2013                 }
2014
2015                 wfProfileOut( $fname );
2016                 return $ret;
2017         }
2018
2019         /**#@+
2020          * Used by doBlockLevels()
2021          * @private
2022          */
2023         /* private */ function closeParagraph() {
2024                 $result = '';
2025                 if ( '' != $this->mLastSection ) {
2026                         $result = '</' . $this->mLastSection  . ">\n";
2027                 }
2028                 $this->mInPre = false;
2029                 $this->mLastSection = '';
2030                 return $result;
2031         }
2032         # getCommon() returns the length of the longest common substring
2033         # of both arguments, starting at the beginning of both.
2034         #
2035         /* private */ function getCommon( $st1, $st2 ) {
2036                 $fl = strlen( $st1 );
2037                 $shorter = strlen( $st2 );
2038                 if ( $fl < $shorter ) { $shorter = $fl; }
2039
2040                 for ( $i = 0; $i < $shorter; ++$i ) {
2041                         if ( $st1{$i} != $st2{$i} ) { break; }
2042                 }
2043                 return $i;
2044         }
2045         # These next three functions open, continue, and close the list
2046         # element appropriate to the prefix character passed into them.
2047         #
2048         /* private */ function openList( $char ) {
2049                 $result = $this->closeParagraph();
2050
2051                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2052                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2053                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2054                 else if ( ';' == $char ) {
2055                         $result .= '<dl><dt>';
2056                         $this->mDTopen = true;
2057                 }
2058                 else { $result = '<!-- ERR 1 -->'; }
2059
2060                 return $result;
2061         }
2062
2063         /* private */ function nextItem( $char ) {
2064                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2065                 else if ( ':' == $char || ';' == $char ) {
2066                         $close = '</dd>';
2067                         if ( $this->mDTopen ) { $close = '</dt>'; }
2068                         if ( ';' == $char ) {
2069                                 $this->mDTopen = true;
2070                                 return $close . '<dt>';
2071                         } else {
2072                                 $this->mDTopen = false;
2073                                 return $close . '<dd>';
2074                         }
2075                 }
2076                 return '<!-- ERR 2 -->';
2077         }
2078
2079         /* private */ function closeList( $char ) {
2080                 if ( '*' == $char ) { $text = '</li></ul>'; }
2081                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2082                 else if ( ':' == $char ) {
2083                         if ( $this->mDTopen ) {
2084                                 $this->mDTopen = false;
2085                                 $text = '</dt></dl>';
2086                         } else {
2087                                 $text = '</dd></dl>';
2088                         }
2089                 }
2090                 else {  return '<!-- ERR 3 -->'; }
2091                 return $text."\n";
2092         }
2093         /**#@-*/
2094
2095         /**
2096          * Make lists from lines starting with ':', '*', '#', etc.
2097          *
2098          * @private
2099          * @return string the lists rendered as HTML
2100          */
2101         function doBlockLevels( $text, $linestart ) {
2102                 $fname = 'Parser::doBlockLevels';
2103                 wfProfileIn( $fname );
2104
2105                 # Parsing through the text line by line.  The main thing
2106                 # happening here is handling of block-level elements p, pre,
2107                 # and making lists from lines starting with * # : etc.
2108                 #
2109                 $textLines = explode( "\n", $text );
2110
2111                 $lastPrefix = $output = '';
2112                 $this->mDTopen = $inBlockElem = false;
2113                 $prefixLength = 0;
2114                 $paragraphStack = false;
2115
2116                 if ( !$linestart ) {
2117                         $output .= array_shift( $textLines );
2118                 }
2119                 foreach ( $textLines as $oLine ) {
2120                         $lastPrefixLength = strlen( $lastPrefix );
2121                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2122                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2123                         if ( !$this->mInPre ) {
2124                                 # Multiple prefixes may abut each other for nested lists.
2125                                 $prefixLength = strspn( $oLine, '*#:;' );
2126                                 $pref = substr( $oLine, 0, $prefixLength );
2127
2128                                 # eh?
2129                                 $pref2 = str_replace( ';', ':', $pref );
2130                                 $t = substr( $oLine, $prefixLength );
2131                                 $this->mInPre = !empty($preOpenMatch);
2132                         } else {
2133                                 # Don't interpret any other prefixes in preformatted text
2134                                 $prefixLength = 0;
2135                                 $pref = $pref2 = '';
2136                                 $t = $oLine;
2137                         }
2138
2139                         # List generation
2140                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2141                                 # Same as the last item, so no need to deal with nesting or opening stuff
2142                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2143                                 $paragraphStack = false;
2144
2145                                 if ( substr( $pref, -1 ) == ';') {
2146                                         # The one nasty exception: definition lists work like this:
2147                                         # ; title : definition text
2148                                         # So we check for : in the remainder text to split up the
2149                                         # title and definition, without b0rking links.
2150                                         $term = $t2 = '';
2151                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2152                                                 $t = $t2;
2153                                                 $output .= $term . $this->nextItem( ':' );
2154                                         }
2155                                 }
2156                         } elseif( $prefixLength || $lastPrefixLength ) {
2157                                 # Either open or close a level...
2158                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2159                                 $paragraphStack = false;
2160
2161                                 while( $commonPrefixLength < $lastPrefixLength ) {
2162                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2163                                         --$lastPrefixLength;
2164                                 }
2165                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2166                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2167                                 }
2168                                 while ( $prefixLength > $commonPrefixLength ) {
2169                                         $char = substr( $pref, $commonPrefixLength, 1 );
2170                                         $output .= $this->openList( $char );
2171
2172                                         if ( ';' == $char ) {
2173                                                 # FIXME: This is dupe of code above
2174                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2175                                                         $t = $t2;
2176                                                         $output .= $term . $this->nextItem( ':' );
2177                                                 }
2178                                         }
2179                                         ++$commonPrefixLength;
2180                                 }
2181                                 $lastPrefix = $pref2;
2182                         }
2183                         if( 0 == $prefixLength ) {
2184                                 wfProfileIn( "$fname-paragraph" );
2185                                 # No prefix (not in list)--go to paragraph mode
2186                                 // XXX: use a stack for nestable elements like span, table and div
2187                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2188                                 $closematch = preg_match(
2189                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2190                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2191                                 if ( $openmatch or $closematch ) {
2192                                         $paragraphStack = false;
2193                                         # TODO bug 5718: paragraph closed
2194                                         $output .= $this->closeParagraph();
2195                                         if ( $preOpenMatch and !$preCloseMatch ) {
2196                                                 $this->mInPre = true;
2197                                         }
2198                                         if ( $closematch ) {
2199                                                 $inBlockElem = false;
2200                                         } else {
2201                                                 $inBlockElem = true;
2202                                         }
2203                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2204                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2205                                                 // pre
2206                                                 if ($this->mLastSection != 'pre') {
2207                                                         $paragraphStack = false;
2208                                                         $output .= $this->closeParagraph().'<pre>';
2209                                                         $this->mLastSection = 'pre';
2210                                                 }
2211                                                 $t = substr( $t, 1 );
2212                                         } else {
2213                                                 // paragraph
2214                                                 if ( '' == trim($t) ) {
2215                                                         if ( $paragraphStack ) {
2216                                                                 $output .= $paragraphStack.'<br />';
2217                                                                 $paragraphStack = false;
2218                                                                 $this->mLastSection = 'p';
2219                                                         } else {
2220                                                                 if ($this->mLastSection != 'p' ) {
2221                                                                         $output .= $this->closeParagraph();
2222                                                                         $this->mLastSection = '';
2223                                                                         $paragraphStack = '<p>';
2224                                                                 } else {
2225                                                                         $paragraphStack = '</p><p>';
2226                                                                 }
2227                                                         }
2228                                                 } else {
2229                                                         if ( $paragraphStack ) {
2230                                                                 $output .= $paragraphStack;
2231                                                                 $paragraphStack = false;
2232                                                                 $this->mLastSection = 'p';
2233                                                         } else if ($this->mLastSection != 'p') {
2234                                                                 $output .= $this->closeParagraph().'<p>';
2235                                                                 $this->mLastSection = 'p';
2236                                                         }
2237                                                 }
2238                                         }
2239                                 }
2240                                 wfProfileOut( "$fname-paragraph" );
2241                         }
2242                         // somewhere above we forget to get out of pre block (bug 785)
2243                         if($preCloseMatch && $this->mInPre) {
2244                                 $this->mInPre = false;
2245                         }
2246                         if ($paragraphStack === false) {
2247                                 $output .= $t."\n";
2248                         }
2249                 }
2250                 while ( $prefixLength ) {
2251                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2252                         --$prefixLength;
2253                 }
2254                 if ( '' != $this->mLastSection ) {
2255                         $output .= '</' . $this->mLastSection . '>';
2256                         $this->mLastSection = '';
2257                 }
2258
2259                 wfProfileOut( $fname );
2260                 return $output;
2261         }
2262
2263         /**
2264          * Split up a string on ':', ignoring any occurences inside tags
2265          * to prevent illegal overlapping.
2266          * @param string $str the string to split
2267          * @param string &$before set to everything before the ':'
2268          * @param string &$after set to everything after the ':'
2269          * return string the position of the ':', or false if none found
2270          */
2271         function findColonNoLinks($str, &$before, &$after) {
2272                 $fname = 'Parser::findColonNoLinks';
2273                 wfProfileIn( $fname );
2274
2275                 $pos = strpos( $str, ':' );
2276                 if( $pos === false ) {
2277                         // Nothing to find!
2278                         wfProfileOut( $fname );
2279                         return false;
2280                 }
2281
2282                 $lt = strpos( $str, '<' );
2283                 if( $lt === false || $lt > $pos ) {
2284                         // Easy; no tag nesting to worry about
2285                         $before = substr( $str, 0, $pos );
2286                         $after = substr( $str, $pos+1 );
2287                         wfProfileOut( $fname );
2288                         return $pos;
2289                 }
2290
2291                 // Ugly state machine to walk through avoiding tags.
2292                 $state = MW_COLON_STATE_TEXT;
2293                 $stack = 0;
2294                 $len = strlen( $str );
2295                 for( $i = 0; $i < $len; $i++ ) {
2296                         $c = $str{$i};
2297
2298                         switch( $state ) {
2299                         // (Using the number is a performance hack for common cases)
2300                         case 0: // MW_COLON_STATE_TEXT:
2301                                 switch( $c ) {
2302                                 case "<":
2303                                         // Could be either a <start> tag or an </end> tag
2304                                         $state = MW_COLON_STATE_TAGSTART;
2305                                         break;
2306                                 case ":":
2307                                         if( $stack == 0 ) {
2308                                                 // We found it!
2309                                                 $before = substr( $str, 0, $i );
2310                                                 $after = substr( $str, $i + 1 );
2311                                                 wfProfileOut( $fname );
2312                                                 return $i;
2313                                         }
2314                                         // Embedded in a tag; don't break it.
2315                                         break;
2316                                 default:
2317                                         // Skip ahead looking for something interesting
2318                                         $colon = strpos( $str, ':', $i );
2319                                         if( $colon === false ) {
2320                                                 // Nothing else interesting
2321                                                 wfProfileOut( $fname );
2322                                                 return false;
2323                                         }
2324                                         $lt = strpos( $str, '<', $i );
2325                                         if( $stack === 0 ) {
2326                                                 if( $lt === false || $colon < $lt ) {
2327                                                         // We found it!
2328                                                         $before = substr( $str, 0, $colon );
2329                                                         $after = substr( $str, $colon + 1 );
2330                                                         wfProfileOut( $fname );
2331                                                         return $i;
2332                                                 }
2333                                         }
2334                                         if( $lt === false ) {
2335                                                 // Nothing else interesting to find; abort!
2336                                                 // We're nested, but there's no close tags left. Abort!
2337                                                 break 2;
2338                                         }
2339                                         // Skip ahead to next tag start
2340                                         $i = $lt;
2341                                         $state = MW_COLON_STATE_TAGSTART;
2342                                 }
2343                                 break;
2344                         case 1: // MW_COLON_STATE_TAG:
2345                                 // In a <tag>
2346                                 switch( $c ) {
2347                                 case ">":
2348                                         $stack++;
2349                                         $state = MW_COLON_STATE_TEXT;
2350                                         break;
2351                                 case "/":
2352                                         // Slash may be followed by >?
2353                                         $state = MW_COLON_STATE_TAGSLASH;
2354                                         break;
2355                                 default:
2356                                         // ignore
2357                                 }
2358                                 break;
2359                         case 2: // MW_COLON_STATE_TAGSTART:
2360                                 switch( $c ) {
2361                                 case "/":
2362                                         $state = MW_COLON_STATE_CLOSETAG;
2363                                         break;
2364                                 case "!":
2365                                         $state = MW_COLON_STATE_COMMENT;
2366                                         break;
2367                                 case ">":
2368                                         // Illegal early close? This shouldn't happen D:
2369                                         $state = MW_COLON_STATE_TEXT;
2370                                         break;
2371                                 default:
2372                                         $state = MW_COLON_STATE_TAG;
2373                                 }
2374                                 break;
2375                         case 3: // MW_COLON_STATE_CLOSETAG:
2376                                 // In a </tag>
2377                                 if( $c == ">" ) {
2378                                         $stack--;
2379                                         if( $stack < 0 ) {
2380                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2381                                                 wfProfileOut( $fname );
2382                                                 return false;
2383                                         }
2384                                         $state = MW_COLON_STATE_TEXT;
2385                                 }
2386                                 break;
2387                         case MW_COLON_STATE_TAGSLASH:
2388                                 if( $c == ">" ) {
2389                                         // Yes, a self-closed tag <blah/>
2390                                         $state = MW_COLON_STATE_TEXT;
2391                                 } else {
2392                                         // Probably we're jumping the gun, and this is an attribute
2393                                         $state = MW_COLON_STATE_TAG;
2394                                 }
2395                                 break;
2396                         case 5: // MW_COLON_STATE_COMMENT:
2397                                 if( $c == "-" ) {
2398                                         $state = MW_COLON_STATE_COMMENTDASH;
2399                                 }
2400                                 break;
2401                         case MW_COLON_STATE_COMMENTDASH:
2402                                 if( $c == "-" ) {
2403                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2404                                 } else {
2405                                         $state = MW_COLON_STATE_COMMENT;
2406                                 }
2407                                 break;
2408                         case MW_COLON_STATE_COMMENTDASHDASH:
2409                                 if( $c == ">" ) {
2410                                         $state = MW_COLON_STATE_TEXT;
2411                                 } else {
2412                                         $state = MW_COLON_STATE_COMMENT;
2413                                 }
2414                                 break;
2415                         default:
2416                                 throw new MWException( "State machine error in $fname" );
2417                         }
2418                 }
2419                 if( $stack > 0 ) {
2420                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2421                         return false;
2422                 }
2423                 wfProfileOut( $fname );
2424                 return false;
2425         }
2426
2427         /**
2428          * Return value of a magic variable (like PAGENAME)
2429          *
2430          * @private
2431          */
2432         function getVariableValue( $index ) {
2433                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2434
2435                 /**
2436                  * Some of these require message or data lookups and can be
2437                  * expensive to check many times.
2438                  */
2439                 static $varCache = array();
2440                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2441                         if ( isset( $varCache[$index] ) ) {
2442                                 return $varCache[$index];
2443                         }
2444                 }
2445
2446                 $ts = time();
2447                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2448
2449                 # Use the time zone
2450                 global $wgLocaltimezone;
2451                 if ( isset( $wgLocaltimezone ) ) {
2452                         $oldtz = getenv( 'TZ' );
2453                         putenv( 'TZ='.$wgLocaltimezone );
2454                 }
2455
2456                 wfSuppressWarnings(); // E_STRICT system time bitching
2457                 $localTimestamp = date( 'YmdHis', $ts );
2458                 $localMonth = date( 'm', $ts );
2459                 $localMonthName = date( 'n', $ts );
2460                 $localDay = date( 'j', $ts );
2461                 $localDay2 = date( 'd', $ts );
2462                 $localDayOfWeek = date( 'w', $ts );
2463                 $localWeek = date( 'W', $ts );
2464                 $localYear = date( 'Y', $ts );
2465                 $localHour = date( 'H', $ts );
2466                 if ( isset( $wgLocaltimezone ) ) {
2467                         putenv( 'TZ='.$oldtz );
2468                 }
2469                 wfRestoreWarnings();
2470
2471                 switch ( $index ) {
2472                         case 'currentmonth':
2473                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2474                         case 'currentmonthname':
2475                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2476                         case 'currentmonthnamegen':
2477                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2478                         case 'currentmonthabbrev':
2479                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2480                         case 'currentday':
2481                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2482                         case 'currentday2':
2483                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2484                         case 'localmonth':
2485                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2486                         case 'localmonthname':
2487                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2488                         case 'localmonthnamegen':
2489                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2490                         case 'localmonthabbrev':
2491                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2492                         case 'localday':
2493                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2494                         case 'localday2':
2495                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2496                         case 'pagename':
2497                                 return wfEscapeWikiText( $this->mTitle->getText() );
2498                         case 'pagenamee':
2499                                 return $this->mTitle->getPartialURL();
2500                         case 'fullpagename':
2501                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2502                         case 'fullpagenamee':
2503                                 return $this->mTitle->getPrefixedURL();
2504                         case 'subpagename':
2505                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2506                         case 'subpagenamee':
2507                                 return $this->mTitle->getSubpageUrlForm();
2508                         case 'basepagename':
2509                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2510                         case 'basepagenamee':
2511                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2512                         case 'talkpagename':
2513                                 if( $this->mTitle->canTalk() ) {
2514                                         $talkPage = $this->mTitle->getTalkPage();
2515                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2516                                 } else {
2517                                         return '';
2518                                 }
2519                         case 'talkpagenamee':
2520                                 if( $this->mTitle->canTalk() ) {
2521                                         $talkPage = $this->mTitle->getTalkPage();
2522                                         return $talkPage->getPrefixedUrl();
2523                                 } else {
2524                                         return '';
2525                                 }
2526                         case 'subjectpagename':
2527                                 $subjPage = $this->mTitle->getSubjectPage();
2528                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2529                         case 'subjectpagenamee':
2530                                 $subjPage = $this->mTitle->getSubjectPage();
2531                                 return $subjPage->getPrefixedUrl();
2532                         case 'revisionid':
2533                                 return $this->mRevisionId;
2534                         case 'revisionday':
2535                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2536                         case 'revisionday2':
2537                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2538                         case 'revisionmonth':
2539                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2540                         case 'revisionyear':
2541                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2542                         case 'revisiontimestamp':
2543                                 return $this->getRevisionTimestamp();
2544                         case 'namespace':
2545                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2546                         case 'namespacee':
2547                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2548                         case 'talkspace':
2549                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2550                         case 'talkspacee':
2551                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2552                         case 'subjectspace':
2553                                 return $this->mTitle->getSubjectNsText();
2554                         case 'subjectspacee':
2555                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2556                         case 'currentdayname':
2557                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2558                         case 'currentyear':
2559                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2560                         case 'currenttime':
2561                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2562                         case 'currenthour':
2563                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2564                         case 'currentweek':
2565                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2566                                 // int to remove the padding
2567                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2568                         case 'currentdow':
2569                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2570                         case 'localdayname':
2571                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2572                         case 'localyear':
2573                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2574                         case 'localtime':
2575                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2576                         case 'localhour':
2577                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2578                         case 'localweek':
2579                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2580                                 // int to remove the padding
2581                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2582                         case 'localdow':
2583                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2584                         case 'numberofarticles':
2585                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2586                         case 'numberoffiles':
2587                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2588                         case 'numberofusers':
2589                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2590                         case 'numberofpages':
2591                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2592                         case 'numberofadmins':
2593                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2594                         case 'numberofedits':
2595                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2596                         case 'currenttimestamp':
2597                                 return $varCache[$index] = wfTimestampNow();
2598                         case 'localtimestamp':
2599                                 return $varCache[$index] = $localTimestamp;
2600                         case 'currentversion':
2601                                 return $varCache[$index] = SpecialVersion::getVersion();
2602                         case 'sitename':
2603                                 return $wgSitename;
2604                         case 'server':
2605                                 return $wgServer;
2606                         case 'servername':
2607                                 return $wgServerName;
2608                         case 'scriptpath':
2609                                 return $wgScriptPath;
2610                         case 'directionmark':
2611                                 return $wgContLang->getDirMark();
2612                         case 'contentlanguage':
2613                                 global $wgContLanguageCode;
2614                                 return $wgContLanguageCode;
2615                         default:
2616                                 $ret = null;
2617                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2618                                         return $ret;
2619                                 else
2620                                         return null;
2621                 }
2622         }
2623
2624         /**
2625          * initialise the magic variables (like CURRENTMONTHNAME)
2626          *
2627          * @private
2628          */
2629         function initialiseVariables() {
2630                 $fname = 'Parser::initialiseVariables';
2631                 wfProfileIn( $fname );
2632                 $variableIDs = MagicWord::getVariableIDs();
2633
2634                 $this->mVariables = array();
2635                 foreach ( $variableIDs as $id ) {
2636                         $mw =& MagicWord::get( $id );
2637                         $mw->addToArray( $this->mVariables, $id );
2638                 }
2639                 wfProfileOut( $fname );
2640         }
2641
2642         /**
2643          * parse any parentheses in format ((title|part|part))
2644          * and call callbacks to get a replacement text for any found piece
2645          *
2646          * @param string $text The text to parse
2647          * @param array $callbacks rules in form:
2648          *     '{' => array(                            # opening parentheses
2649          *                                      'end' => '}',   # closing parentheses
2650          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2651          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2652          *                                                                )
2653          *                                      )
2654          *                                      'min' => 2,     # Minimum parenthesis count in cb
2655          *                                      'max' => 3,     # Maximum parenthesis count in cb
2656          * @private
2657          */
2658         function replace_callback ($text, $callbacks) {
2659                 wfProfileIn( __METHOD__ );
2660                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2661                 $lastOpeningBrace = -1;                 # last not closed parentheses
2662
2663                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2664
2665                 $i = 0;
2666                 while ( $i < strlen( $text ) ) {
2667                         # Find next opening brace, closing brace or pipe
2668                         if ( $lastOpeningBrace == -1 ) {
2669                                 $currentClosing = '';
2670                                 $search = $validOpeningBraces;
2671                         } else {
2672                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2673                                 $search = $validOpeningBraces . '|' . $currentClosing;
2674                         }
2675                         $rule = null;
2676                         $i += strcspn( $text, $search, $i );
2677                         if ( $i < strlen( $text ) ) {
2678                                 if ( $text[$i] == '|' ) {
2679                                         $found = 'pipe';
2680                                 } elseif ( $text[$i] == $currentClosing ) {
2681                                         $found = 'close';
2682                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2683                                         $found = 'open';
2684                                         $rule = $callbacks[$text[$i]];
2685                                 } else {
2686                                         # Some versions of PHP have a strcspn which stops on null characters
2687                                         # Ignore and continue
2688                                         ++$i;
2689                                         continue;
2690                                 }
2691                         } else {
2692                                 # All done
2693                                 break;
2694                         }
2695
2696                         if ( $found == 'open' ) {
2697                                 # found opening brace, let's add it to parentheses stack
2698                                 $piece = array('brace' => $text[$i],
2699                                                            'braceEnd' => $rule['end'],
2700                                                            'title' => '',
2701                                                            'parts' => null);
2702
2703                                 # count opening brace characters
2704                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2705                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2706                                 $i += $piece['count'];
2707
2708                                 # we need to add to stack only if opening brace count is enough for one of the rules
2709                                 if ( $piece['count'] >= $rule['min'] ) {
2710                                         $lastOpeningBrace ++;
2711                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2712                                 }
2713                         } elseif ( $found == 'close' ) {
2714                                 # lets check if it is enough characters for closing brace
2715                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2716                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2717
2718                                 # check for maximum matching characters (if there are 5 closing
2719                                 # characters, we will probably need only 3 - depending on the rules)
2720                                 $matchingCount = 0;
2721                                 $matchingCallback = null;
2722                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2723                                 if ( $count > $cbType['max'] ) {
2724                                         # The specified maximum exists in the callback array, unless the caller
2725                                         # has made an error
2726                                         $matchingCount = $cbType['max'];
2727                                 } else {
2728                                         # Count is less than the maximum
2729                                         # Skip any gaps in the callback array to find the true largest match
2730                                         # Need to use array_key_exists not isset because the callback can be null
2731                                         $matchingCount = $count;
2732                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2733                                                 --$matchingCount;
2734                                         }
2735                                 }
2736
2737                                 if ($matchingCount <= 0) {
2738                                         $i += $count;
2739                                         continue;
2740                                 }
2741                                 $matchingCallback = $cbType['cb'][$matchingCount];
2742
2743                                 # let's set a title or last part (if '|' was found)
2744                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2745                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2746                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2747                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2748                                 } else {
2749                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2750                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2751                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2752                                 }
2753
2754                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2755                                 $pieceEnd = $i + $matchingCount;
2756
2757                                 if( is_callable( $matchingCallback ) ) {
2758                                         $cbArgs = array (
2759                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2760                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2761                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2762                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2763                                                                          );
2764                                         # finally we can call a user callback and replace piece of text
2765                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2766                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2767                                         $i = $pieceStart + strlen($replaceWith);
2768                                 } else {
2769                                         # null value for callback means that parentheses should be parsed, but not replaced
2770                                         $i += $matchingCount;
2771                                 }
2772
2773                                 # reset last opening parentheses, but keep it in case there are unused characters
2774                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2775                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2776                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2777                                                            'title' => '',
2778                                                            'parts' => null,
2779                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2780                                 $openingBraceStack[$lastOpeningBrace--] = null;
2781
2782                                 if ($matchingCount < $piece['count']) {
2783                                         $piece['count'] -= $matchingCount;
2784                                         $piece['startAt'] -= $matchingCount;
2785                                         $piece['partStart'] = $piece['startAt'];
2786                                         # do we still qualify for any callback with remaining count?
2787                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2788                                         while ( $piece['count'] ) {
2789                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2790                                                         $lastOpeningBrace++;
2791                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2792                                                         break;
2793                                                 }
2794                                                 --$piece['count'];
2795                                         }
2796                                 }
2797                         } elseif ( $found == 'pipe' ) {
2798                                 # lets set a title if it is a first separator, or next part otherwise
2799                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2800                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2801                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2802                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2803                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2804                                 } else {
2805                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2806                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2807                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2808                                 }
2809                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2810                         }
2811                 }
2812
2813                 wfProfileOut( __METHOD__ );
2814                 return $text;
2815         }
2816
2817         /**
2818          * Replace magic variables, templates, and template arguments
2819          * with the appropriate text. Templates are substituted recursively,
2820          * taking care to avoid infinite loops.
2821          *
2822          * Note that the substitution depends on value of $mOutputType:
2823          *  OT_WIKI: only {{subst:}} templates
2824          *  OT_MSG: only magic variables
2825          *  OT_HTML: all templates and magic variables
2826          *
2827          * @param string $tex The text to transform
2828          * @param array $args Key-value pairs representing template parameters to substitute
2829          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2830          * @private
2831          */
2832         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2833                 # Prevent too big inclusions
2834                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2835                         return $text;
2836                 }
2837
2838                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2839                 wfProfileIn( $fname );
2840
2841                 # This function is called recursively. To keep track of arguments we need a stack:
2842                 array_push( $this->mArgStack, $args );
2843
2844                 $braceCallbacks = array();
2845                 if ( !$argsOnly ) {
2846                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2847                 }
2848                 if ( $this->mOutputType != OT_MSG ) {
2849                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2850                 }
2851                 if ( $braceCallbacks ) {
2852                         $callbacks = array(
2853                                 '{' => array(
2854                                         'end' => '}',
2855                                         'cb' => $braceCallbacks,
2856                                         'min' => $argsOnly ? 3 : 2,
2857                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2858                                 ),
2859                                 '[' => array(
2860                                         'end' => ']',
2861                                         'cb' => array(2=>null),
2862                                         'min' => 2,
2863                                         'max' => 2,
2864                                 )
2865                         );
2866                         $text = $this->replace_callback ($text, $callbacks);
2867
2868                         array_pop( $this->mArgStack );
2869                 }
2870                 wfProfileOut( $fname );
2871                 return $text;
2872         }
2873
2874         /**
2875          * Replace magic variables
2876          * @private
2877          */
2878         function variableSubstitution( $matches ) {
2879                 global $wgContLang;
2880                 $fname = 'Parser::variableSubstitution';
2881                 $varname = $wgContLang->lc($matches[1]);
2882                 wfProfileIn( $fname );
2883                 $skip = false;
2884                 if ( $this->mOutputType == OT_WIKI ) {
2885                         # Do only magic variables prefixed by SUBST
2886                         $mwSubst =& MagicWord::get( 'subst' );
2887                         if (!$mwSubst->matchStartAndRemove( $varname ))
2888                                 $skip = true;
2889                         # Note that if we don't substitute the variable below,
2890                         # we don't remove the {{subst:}} magic word, in case
2891                         # it is a template rather than a magic variable.
2892                 }
2893                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2894                         $id = $this->mVariables[$varname];
2895                         # Now check if we did really match, case sensitive or not
2896                         $mw =& MagicWord::get( $id );
2897                         if ($mw->match($matches[1])) {
2898                                 $text = $this->getVariableValue( $id );
2899                                 $this->mOutput->mContainsOldMagic = true;
2900                         } else {
2901                                 $text = $matches[0];
2902                         }
2903                 } else {
2904                         $text = $matches[0];
2905                 }
2906                 wfProfileOut( $fname );
2907                 return $text;
2908         }
2909
2910
2911         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2912         static function createAssocArgs( $args ) {
2913                 $assocArgs = array();
2914                 $index = 1;
2915                 foreach( $args as $arg ) {
2916                         $eqpos = strpos( $arg, '=' );
2917                         if ( $eqpos === false ) {
2918                                 $assocArgs[$index++] = $arg;
2919                         } else {
2920                                 $name = trim( substr( $arg, 0, $eqpos ) );
2921                                 $value = trim( substr( $arg, $eqpos+1 ) );
2922                                 if ( $value === false ) {
2923                                         $value = '';
2924                                 }
2925                                 if ( $name !== false ) {
2926                                         $assocArgs[$name] = $value;
2927                                 }
2928                         }
2929                 }
2930
2931                 return $assocArgs;
2932         }
2933
2934         /**
2935          * Return the text of a template, after recursively
2936          * replacing any variables or templates within the template.
2937          *
2938          * @param array $piece The parts of the template
2939          *  $piece['text']: matched text
2940          *  $piece['title']: the title, i.e. the part before the |
2941          *  $piece['parts']: the parameter array
2942          * @return string the text of the template
2943          * @private
2944          */
2945         function braceSubstitution( $piece ) {
2946                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2947                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2948                 wfProfileIn( $fname );
2949                 wfProfileIn( __METHOD__.'-setup' );
2950
2951                 # Flags
2952                 $found = false;             # $text has been filled
2953                 $nowiki = false;            # wiki markup in $text should be escaped
2954                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2955                 $noargs = false;            # Don't replace triple-brace arguments in $text
2956                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2957                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2958                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2959                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2960
2961                 # Title object, where $text came from
2962                 $title = NULL;
2963
2964                 $linestart = '';
2965
2966
2967                 # $part1 is the bit before the first |, and must contain only title characters
2968                 # $args is a list of arguments, starting from index 0, not including $part1
2969
2970                 $titleText = $part1 = $piece['title'];
2971                 # If the third subpattern matched anything, it will start with |
2972
2973                 if (null == $piece['parts']) {
2974                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2975                         if ($replaceWith != $piece['text']) {
2976                                 $text = $replaceWith;
2977                                 $found = true;
2978                                 $noparse = true;
2979                                 $noargs = true;
2980                         }
2981                 }
2982
2983                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2984                 wfProfileOut( __METHOD__.'-setup' );
2985
2986                 # SUBST
2987                 wfProfileIn( __METHOD__.'-modifiers' );
2988                 if ( !$found ) {
2989                         $mwSubst =& MagicWord::get( 'subst' );
2990                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2991                                 # One of two possibilities is true:
2992                                 # 1) Found SUBST but not in the PST phase
2993                                 # 2) Didn't find SUBST and in the PST phase
2994                                 # In either case, return without further processing
2995                                 $text = $piece['text'];
2996                                 $found = true;
2997                                 $noparse = true;
2998                                 $noargs = true;
2999                         }
3000                 }
3001
3002                 # MSG, MSGNW and RAW
3003                 if ( !$found ) {
3004                         # Check for MSGNW:
3005                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3006                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3007                                 $nowiki = true;
3008                         } else {
3009                                 # Remove obsolete MSG:
3010                                 $mwMsg =& MagicWord::get( 'msg' );
3011                                 $mwMsg->matchStartAndRemove( $part1 );
3012                         }
3013
3014                         # Check for RAW:
3015                         $mwRaw =& MagicWord::get( 'raw' );
3016                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3017                                 $forceRawInterwiki = true;
3018                         }
3019                 }
3020                 wfProfileOut( __METHOD__.'-modifiers' );
3021
3022                 //save path level before recursing into functions & templates.
3023                 $lastPathLevel = $this->mTemplatePath;
3024
3025                 # Parser functions
3026                 if ( !$found ) {
3027                         wfProfileIn( __METHOD__ . '-pfunc' );
3028
3029                         $colonPos = strpos( $part1, ':' );
3030                         if ( $colonPos !== false ) {
3031                                 # Case sensitive functions
3032                                 $function = substr( $part1, 0, $colonPos );
3033                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3034                                         $function = $this->mFunctionSynonyms[1][$function];
3035                                 } else {
3036                                         # Case insensitive functions
3037                                         $function = strtolower( $function );
3038                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3039                                                 $function = $this->mFunctionSynonyms[0][$function];
3040                                         } else {
3041                                                 $function = false;
3042                                         }
3043                                 }
3044                                 if ( $function ) {
3045                                         $funcArgs = array_map( 'trim', $args );
3046                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
3047                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
3048                                         $found = true;
3049
3050                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3051                                         //$noargs = true;
3052                                         //$noparse = true;
3053
3054                                         if ( is_array( $result ) ) {
3055                                                 if ( isset( $result[0] ) ) {
3056                                                         $text = $linestart . $result[0];
3057                                                         unset( $result[0] );
3058                                                 }
3059
3060                                                 // Extract flags into the local scope
3061                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3062                                                 extract( $result );
3063                                         } else {
3064                                                 $text = $linestart . $result;
3065                                         }
3066                                 }
3067                         }
3068                         wfProfileOut( __METHOD__ . '-pfunc' );
3069                 }
3070
3071                 # Template table test
3072
3073                 # Did we encounter this template already? If yes, it is in the cache
3074                 # and we need to check for loops.
3075                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3076                         $found = true;
3077
3078                         # Infinite loop test
3079                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3080                                 $noparse = true;
3081                                 $noargs = true;
3082                                 $found = true;
3083                                 $text = $linestart .
3084                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3085                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3086                         } else {
3087                                 # set $text to cached message.
3088                                 $text = $linestart . $this->mTemplates[$piece['title']];
3089                                 #treat title for cached page the same as others
3090                                 $ns = NS_TEMPLATE;
3091                                 $subpage = '';
3092                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3093                                 if ($subpage !== '') {
3094                                   $ns = $this->mTitle->getNamespace();
3095                                 }
3096                                 $title = Title::newFromText( $part1, $ns );
3097                                 //used by include size checking
3098                                 $titleText = $title->getPrefixedText();
3099                                 //used by edit section links
3100                                 $replaceHeadings = true;
3101
3102                         }
3103                 }
3104
3105                 # Load from database
3106                 if ( !$found ) {
3107                         wfProfileIn( __METHOD__ . '-loadtpl' );
3108                         $ns = NS_TEMPLATE;
3109                         # declaring $subpage directly in the function call
3110                         # does not work correctly with references and breaks
3111                         # {{/subpage}}-style inclusions
3112                         $subpage = '';
3113                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3114                         if ($subpage !== '') {
3115                                 $ns = $this->mTitle->getNamespace();
3116                         }
3117                         $title = Title::newFromText( $part1, $ns );
3118
3119
3120                         if ( !is_null( $title ) ) {
3121                                 $titleText = $title->getPrefixedText();
3122                                 # Check for language variants if the template is not found
3123                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3124                                         $wgContLang->findVariantLink($part1, $title);
3125                                 }
3126
3127                                 if ( !$title->isExternal() ) {
3128                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3129                                                 $text = SpecialPage::capturePath( $title );
3130                                                 if ( is_string( $text ) ) {
3131                                                         $found = true;
3132                                                         $noparse = true;
3133                                                         $noargs = true;
3134                                                         $isHTML = true;
3135                                                         $this->disableCache();
3136                                                 }
3137                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3138                                                 $found = false; //access denied
3139                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3140                                         } else {
3141                                                 list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3142                                                 if ( $articleContent !== false ) {
3143                                                         $found = true;
3144                                                         $text = $articleContent;
3145                                                         $replaceHeadings = true;
3146                                                 }
3147                                         }
3148
3149                                         # If the title is valid but undisplayable, make a link to it
3150                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3151                                                 $text = "[[:$titleText]]";
3152                                                 $found = true;
3153                                         }
3154                                 } elseif ( $title->isTrans() ) {
3155                                         // Interwiki transclusion
3156                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3157                                                 $text = $this->interwikiTransclude( $title, 'render' );
3158                                                 $isHTML = true;
3159                                                 $noparse = true;
3160                                         } else {
3161                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3162                                                 $replaceHeadings = true;
3163                                         }
3164                                         $found = true;
3165                                 }
3166
3167                                 # Template cache array insertion
3168                                 # Use the original $piece['title'] not the mangled $part1, so that
3169                                 # modifiers such as RAW: produce separate cache entries
3170                                 if( $found ) {
3171                                         if( $isHTML ) {
3172                                                 // A special page; don't store it in the template cache.
3173                                         } else {
3174                                                 $this->mTemplates[$piece['title']] = $text;
3175                                         }
3176                                         $text = $linestart . $text;
3177                                 }
3178                         }
3179                         wfProfileOut( __METHOD__ . '-loadtpl' );
3180                 }
3181
3182                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3183                         # Error, oversize inclusion
3184                         $text = $linestart .
3185                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3186                         $noparse = true;
3187                         $noargs = true;
3188                 }
3189
3190                 # Recursive parsing, escaping and link table handling
3191                 # Only for HTML output
3192                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3193                         $text = wfEscapeWikiText( $text );
3194                 } elseif ( !$this->ot['msg'] && $found ) {
3195                         if ( $noargs ) {
3196                                 $assocArgs = array();
3197                         } else {
3198                                 # Clean up argument array
3199                                 $assocArgs = self::createAssocArgs($args);
3200                                 # Add a new element to the templace recursion path
3201                                 $this->mTemplatePath[$part1] = 1;
3202                         }
3203
3204                         if ( !$noparse ) {
3205                                 # If there are any <onlyinclude> tags, only include them
3206                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3207                                         $replacer = new OnlyIncludeReplacer;
3208                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3209                                                 array( &$replacer, 'replace' ), $text );
3210                                         $text = $replacer->output;
3211                                 }
3212                                 # Remove <noinclude> sections and <includeonly> tags
3213                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3214                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3215
3216                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3217                                         # Strip <nowiki>, <pre>, etc.
3218                                         $text = $this->strip( $text, $this->mStripState );
3219                                         if ( $this->ot['html'] ) {
3220                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3221                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3222                                                 $text = Sanitizer::removeHTMLcomments( $text );
3223                                         }
3224                                 }
3225                                 $text = $this->replaceVariables( $text, $assocArgs );
3226
3227                                 # If the template begins with a table or block-level
3228                                 # element, it should be treated as beginning a new line.
3229                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3230                                         $text = "\n" . $text;
3231                                 }
3232                         } elseif ( !$noargs ) {
3233                                 # $noparse and !$noargs
3234                                 # Just replace the arguments, not any double-brace items
3235                                 # This is used for rendered interwiki transclusion
3236                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3237                         }
3238                 }
3239                 # Prune lower levels off the recursion check path
3240                 $this->mTemplatePath = $lastPathLevel;
3241
3242                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3243                         # Error, oversize inclusion
3244                         $text = $linestart .
3245                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3246                         $noparse = true;
3247                         $noargs = true;
3248                 }
3249
3250                 if ( !$found ) {
3251                         wfProfileOut( $fname );
3252                         return $piece['text'];
3253                 } else {
3254                         wfProfileIn( __METHOD__ . '-placeholders' );
3255                         if ( $isHTML ) {
3256                                 # Replace raw HTML by a placeholder
3257                                 # Add a blank line preceding, to prevent it from mucking up
3258                                 # immediately preceding headings
3259                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3260                         } else {
3261                                 # replace ==section headers==
3262                                 # XXX this needs to go away once we have a better parser.
3263                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3264                                         if( !is_null( $title ) )
3265                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3266                                         else
3267                                                 $encodedname = base64_encode("");
3268                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3269                                                 PREG_SPLIT_DELIM_CAPTURE);
3270                                         $text = '';
3271                                         $nsec = $headingOffset;
3272
3273                                         for( $i = 0; $i < count($m); $i += 2 ) {
3274                                                 $text .= $m[$i];
3275                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3276                                                 $hl = $m[$i + 1];
3277                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3278                                                         $text .= $hl;
3279                                                         continue;
3280                                                 }
3281                                                 $m2 = array();
3282                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3283                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3284                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3285
3286                                                 $nsec++;
3287                                         }
3288                                 }
3289                         }
3290                         wfProfileOut( __METHOD__ . '-placeholders' );
3291                 }
3292
3293                 # Prune lower levels off the recursion check path
3294                 $this->mTemplatePath = $lastPathLevel;
3295
3296                 if ( !$found ) {
3297                         wfProfileOut( $fname );
3298                         return $piece['text'];
3299                 } else {
3300                         wfProfileOut( $fname );
3301                         return $text;
3302                 }
3303         }
3304
3305         /**
3306          * Fetch the unparsed text of a template and register a reference to it.
3307          */
3308         function fetchTemplateAndtitle( $title ) {
3309                 $text = $skip = false;
3310                 $finalTitle = $title;
3311                 // Loop to fetch the article, with up to 1 redirect
3312                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3313                         # Give extensions a chance to select the revision instead
3314                         $id = false; // Assume current
3315                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( &$this, &$title, &$skip, &$id ) );
3316
3317                         if( $skip ) {
3318                                 $text = false;
3319                                 $this->mOutput->addTemplate( $title, $title->getArticleID(), null );
3320                                 break;
3321                         }
3322                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3323                         $rev_id = $rev ? $rev->getId() : 0;
3324
3325                         $this->mOutput->addTemplate( $title, $title->getArticleID(), $rev_id );
3326
3327                         if( $rev ) {
3328                                 $text = $rev->getText();
3329                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3330                                 global $wgLang;
3331                                 $message = $wgLang->lcfirst( $title->getText() );
3332                                 $text = wfMsgForContentNoTrans( $message );
3333                                 if( wfEmptyMsg( $message, $text ) ) {
3334                                         $text = false;
3335                                         break;
3336                                 }
3337                         } else {
3338                                 break;
3339                         }
3340                         if ( $text === false ) {
3341                                 break;
3342                         }
3343                         // Redirect?
3344                         $finalTitle = $title;
3345                         $title = Title::newFromRedirect( $text );
3346                 }
3347                 return array($text,$finalTitle);
3348         }
3349
3350         function fetchTemplate( $title ) {
3351                 $rv = $this->fetchTemplateAndtitle($title);
3352                 return $rv[0];
3353         }
3354
3355         /**
3356          * Transclude an interwiki link.
3357          */
3358         function interwikiTransclude( $title, $action ) {
3359                 global $wgEnableScaryTranscluding;
3360
3361                 if (!$wgEnableScaryTranscluding)
3362                         return wfMsg('scarytranscludedisabled');
3363
3364                 $url = $title->getFullUrl( "action=$action" );
3365
3366                 if (strlen($url) > 255)
3367                         return wfMsg('scarytranscludetoolong');
3368                 return $this->fetchScaryTemplateMaybeFromCache($url);
3369         }
3370
3371         function fetchScaryTemplateMaybeFromCache($url) {
3372                 global $wgTranscludeCacheExpiry;
3373                 $dbr = wfGetDB(DB_SLAVE);
3374                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3375                                 array('tc_url' => $url));
3376                 if ($obj) {
3377                         $time = $obj->tc_time;
3378                         $text = $obj->tc_contents;
3379                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3380                                 return $text;
3381                         }
3382                 }
3383
3384                 $text = Http::get($url);
3385                 if (!$text)
3386                         return wfMsg('scarytranscludefailed', $url);
3387
3388                 $dbw = wfGetDB(DB_MASTER);
3389                 $dbw->replace('transcache', array('tc_url'), array(
3390                         'tc_url' => $url,
3391                         'tc_time' => time(),
3392                         'tc_contents' => $text));
3393                 return $text;
3394         }
3395
3396
3397         /**
3398          * Triple brace replacement -- used for template arguments
3399          * @private
3400          */
3401         function argSubstitution( $matches ) {
3402                 $arg = trim( $matches['title'] );
3403                 $text = $matches['text'];
3404                 $inputArgs = end( $this->mArgStack );
3405
3406                 if ( array_key_exists( $arg, $inputArgs ) ) {
3407                         $text = $inputArgs[$arg];
3408                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3409                 null != $matches['parts'] && count($matches['parts']) > 0) {
3410                         $text = $matches['parts'][0];
3411                 }
3412                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3413                         $text = $matches['text'] .
3414                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3415                 }
3416
3417                 return $text;
3418         }
3419
3420         /**
3421          * Increment an include size counter
3422          *
3423          * @param string $type The type of expansion
3424          * @param integer $size The size of the text
3425          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3426          */
3427         function incrementIncludeSize( $type, $size ) {
3428                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3429                         return false;
3430                 } else {
3431                         $this->mIncludeSizes[$type] += $size;
3432                         return true;
3433                 }
3434         }
3435
3436         /**
3437          * Detect __NOGALLERY__ magic word and set a placeholder
3438          */
3439         function stripNoGallery( &$text ) {
3440                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3441                 # do not add TOC
3442                 $mw = MagicWord::get( 'nogallery' );
3443                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3444         }
3445
3446         /**
3447          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3448          * placeholder that will then be replaced by the real TOC in
3449          * ->formatHeadings, this works because at this points real
3450          * comments will have already been discarded by the sanitizer.
3451          *
3452          * Any additional __TOC__ magic words left over will be discarded
3453          * as there can only be one TOC on the page.
3454          */
3455         function stripToc( $text ) {
3456                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3457                 # do not add TOC
3458                 $mw = MagicWord::get( 'notoc' );
3459                 if( $mw->matchAndRemove( $text ) ) {
3460                         $this->mShowToc = false;
3461                 }
3462
3463                 $mw = MagicWord::get( 'toc' );
3464                 if( $mw->match( $text ) ) {
3465                         $this->mShowToc = true;
3466                         $this->mForceTocPosition = true;
3467
3468                         // Set a placeholder. At the end we'll fill it in with the TOC.
3469                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3470
3471                         // Only keep the first one.
3472                         $text = $mw->replace( '', $text );
3473                 }
3474                 return $text;
3475         }
3476
3477         /**
3478          * This function accomplishes several tasks:
3479          * 1) Auto-number headings if that option is enabled
3480          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3481          * 3) Add a Table of contents on the top for users who have enabled the option
3482          * 4) Auto-anchor headings
3483          *
3484          * It loops through all headlines, collects the necessary data, then splits up the
3485          * string and re-inserts the newly formatted headlines.
3486          *
3487          * @param string $text
3488          * @param boolean $isMain
3489          * @private
3490          */
3491         function formatHeadings( $text, $isMain=true ) {
3492                 global $wgMaxTocLevel, $wgContLang;
3493
3494                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3495                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3496                         $showEditLink = 0;
3497                 } else {
3498                         $showEditLink = $this->mOptions->getEditSection();
3499                 }
3500
3501                 # Inhibit editsection links if requested in the page
3502                 $esw =& MagicWord::get( 'noeditsection' );
3503                 if( $esw->matchAndRemove( $text ) ) {
3504                         $showEditLink = 0;
3505                 }
3506
3507                 # Get all headlines for numbering them and adding funky stuff like [edit]
3508                 # links - this is for later, but we need the number of headlines right now
3509                 $matches = array();
3510                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3511
3512                 # if there are fewer than 4 headlines in the article, do not show TOC
3513                 # unless it's been explicitly enabled.
3514                 $enoughToc = $this->mShowToc &&
3515                         (($numMatches >= 4) || $this->mForceTocPosition);
3516
3517                 # Allow user to stipulate that a page should have a "new section"
3518                 # link added via __NEWSECTIONLINK__
3519                 $mw =& MagicWord::get( 'newsectionlink' );
3520                 if( $mw->matchAndRemove( $text ) )
3521                         $this->mOutput->setNewSection( true );
3522
3523                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3524                 # override above conditions and always show TOC above first header
3525                 $mw =& MagicWord::get( 'forcetoc' );
3526                 if ($mw->matchAndRemove( $text ) ) {
3527                         $this->mShowToc = true;
3528                         $enoughToc = true;
3529                 }
3530
3531                 # We need this to perform operations on the HTML
3532                 $sk = $this->mOptions->getSkin();
3533
3534                 # headline counter
3535                 $headlineCount = 0;
3536                 $sectionCount = 0; # headlineCount excluding template sections
3537                 $numVisible = 0;
3538
3539                 # Ugh .. the TOC should have neat indentation levels which can be
3540                 # passed to the skin functions. These are determined here
3541                 $toc = '';
3542                 $full = '';
3543                 $head = array();
3544                 $sublevelCount = array();
3545                 $levelCount = array();
3546                 $toclevel = 0;
3547                 $level = 0;
3548                 $prevlevel = 0;
3549                 $toclevel = 0;
3550                 $prevtoclevel = 0;
3551
3552                 foreach( $matches[3] as $headline ) {
3553                         $istemplate = 0;
3554                         $templatetitle = '';
3555                         $templatesection = 0;
3556                         $numbering = '';
3557                         $mat = array();
3558                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3559                                 $istemplate = 1;
3560                                 $templatetitle = base64_decode($mat[1]);
3561                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3562                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3563                         }
3564
3565                         if( $toclevel ) {
3566                                 $prevlevel = $level;
3567                                 $prevtoclevel = $toclevel;
3568                         }
3569                         $level = $matches[1][$headlineCount];
3570
3571                         if( $doNumberHeadings || $enoughToc ) {
3572
3573                                 if ( $level > $prevlevel ) {
3574                                         # Increase TOC level
3575                                         $toclevel++;
3576                                         $sublevelCount[$toclevel] = 0;
3577                                         if( $toclevel<$wgMaxTocLevel ) {
3578                                                 $prevtoclevel = $toclevel;
3579                                                 $toc .= $sk->tocIndent();
3580                                                 $numVisible++;
3581                                         }
3582                                 }
3583                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3584                                         # Decrease TOC level, find level to jump to
3585
3586                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3587                                                 # Can only go down to level 1
3588                                                 $toclevel = 1;
3589                                         } else {
3590                                                 for ($i = $toclevel; $i > 0; $i--) {
3591                                                         if ( $levelCount[$i] == $level ) {
3592                                                                 # Found last matching level
3593                                                                 $toclevel = $i;
3594                                                                 break;
3595                                                         }
3596                                                         elseif ( $levelCount[$i] < $level ) {
3597                                                                 # Found first matching level below current level
3598                                                                 $toclevel = $i + 1;
3599                                                                 break;
3600                                                         }
3601                                                 }
3602                                         }
3603                                         if( $toclevel<$wgMaxTocLevel ) {
3604                                                 if($prevtoclevel < $wgMaxTocLevel) {
3605                                                         # Unindent only if the previous toc level was shown :p
3606                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3607                                                 } else {
3608                                                         $toc .= $sk->tocLineEnd();
3609                                                 }
3610                                         }
3611                                 }
3612                                 else {
3613                                         # No change in level, end TOC line
3614                                         if( $toclevel<$wgMaxTocLevel ) {
3615                                                 $toc .= $sk->tocLineEnd();
3616                                         }
3617                                 }
3618
3619                                 $levelCount[$toclevel] = $level;
3620
3621                                 # count number of headlines for each level
3622                                 @$sublevelCount[$toclevel]++;
3623                                 $dot = 0;
3624                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3625                                         if( !empty( $sublevelCount[$i] ) ) {
3626                                                 if( $dot ) {
3627                                                         $numbering .= '.';
3628                                                 }
3629                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3630                                                 $dot = 1;
3631                                         }
3632                                 }
3633                         }
3634
3635                         # The canonized header is a version of the header text safe to use for links
3636                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3637                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3638
3639                         # Remove link placeholders by the link text.
3640                         #     <!--LINK number-->
3641                         # turns into
3642                         #     link text with suffix
3643                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3644                                                             "\$this->mLinkHolders['texts'][\$1]",
3645                                                             $canonized_headline );
3646                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3647                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3648                                                             $canonized_headline );
3649
3650                         # strip out HTML
3651                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3652                         $tocline = trim( $canonized_headline );
3653                         # Save headline for section edit hint before it's escaped
3654                         $headline_hint = trim( $canonized_headline );
3655                         $canonized_headline = Sanitizer::escapeId( $tocline );
3656                         $refers[$headlineCount] = $canonized_headline;
3657
3658                         # count how many in assoc. array so we can track dupes in anchors
3659                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3660                         $refcount[$headlineCount]=$refers[$canonized_headline];
3661
3662                         # Don't number the heading if it is the only one (looks silly)
3663                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3664                                 # the two are different if the line contains a link
3665                                 $headline=$numbering . ' ' . $headline;
3666                         }
3667
3668                         # Create the anchor for linking from the TOC to the section
3669                         $anchor = $canonized_headline;
3670                         if($refcount[$headlineCount] > 1 ) {
3671                                 $anchor .= '_' . $refcount[$headlineCount];
3672                         }
3673                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3674                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3675                         }
3676                         # give headline the correct <h#> tag
3677                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3678                                 if( $istemplate )
3679                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3680                                 else
3681                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3682                         } else {
3683                                 $editlink = '';
3684                         }
3685                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3686
3687                         $headlineCount++;
3688                         if( !$istemplate )
3689                                 $sectionCount++;
3690                 }
3691
3692                 # Never ever show TOC if no headers
3693                 if( $numVisible < 1 ) {
3694                         $enoughToc = false;
3695                 }
3696
3697                 if( $enoughToc ) {
3698                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3699                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3700                         }
3701                         $toc = $sk->tocList( $toc );
3702                 }
3703
3704                 # split up and insert constructed headlines
3705
3706                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3707                 $i = 0;
3708
3709                 foreach( $blocks as $block ) {
3710                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3711                                 # This is the [edit] link that appears for the top block of text when
3712                                 # section editing is enabled
3713
3714                                 # Disabled because it broke block formatting
3715                                 # For example, a bullet point in the top line
3716                                 # $full .= $sk->editSectionLink(0);
3717                         }
3718                         $full .= $block;
3719                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3720                                 # Top anchor now in skin
3721                                 $full = $full.$toc;
3722                         }
3723
3724                         if( !empty( $head[$i] ) ) {
3725                                 $full .= $head[$i];
3726                         }
3727                         $i++;
3728                 }
3729                 if( $this->mForceTocPosition ) {
3730                         return str_replace( '<!--MWTOC-->', $toc, $full );
3731                 } else {
3732                         return $full;
3733                 }
3734         }
3735
3736         /**
3737          * Transform wiki markup when saving a page by doing \r\n -> \n
3738          * conversion, substitting signatures, {{subst:}} templates, etc.
3739          *
3740          * @param string $text the text to transform
3741          * @param Title &$title the Title object for the current article
3742          * @param User &$user the User object describing the current user
3743          * @param ParserOptions $options parsing options
3744          * @param bool $clearState whether to clear the parser state first
3745          * @return string the altered wiki markup
3746          * @public
3747          */
3748         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3749                 $this->mOptions = $options;
3750                 $this->mTitle =& $title;
3751                 $this->setOutputType( OT_WIKI );
3752
3753                 if ( $clearState ) {
3754                         $this->clearState();
3755                 }
3756
3757                 $stripState = new StripState;
3758                 $pairs = array(
3759                         "\r\n" => "\n",
3760                 );
3761                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3762                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3763                 $text = $this->pstPass2( $text, $stripState, $user );
3764                 $text = $stripState->unstripBoth( $text );
3765                 return $text;
3766         }
3767
3768         /**
3769          * Pre-save transform helper function
3770          * @private
3771          */
3772         function pstPass2( $text, &$stripState, $user ) {
3773                 global $wgContLang, $wgLocaltimezone;
3774
3775                 /* Note: This is the timestamp saved as hardcoded wikitext to
3776                  * the database, we use $wgContLang here in order to give
3777                  * everyone the same signature and use the default one rather
3778                  * than the one selected in each user's preferences.
3779                  */
3780                 if ( isset( $wgLocaltimezone ) ) {
3781                         $oldtz = getenv( 'TZ' );
3782                         putenv( 'TZ='.$wgLocaltimezone );
3783                 }
3784                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3785                   ' (' . date( 'T' ) . ')';
3786                 if ( isset( $wgLocaltimezone ) ) {
3787                         putenv( 'TZ='.$oldtz );
3788                 }
3789
3790                 # Variable replacement
3791                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3792                 $text = $this->replaceVariables( $text );
3793
3794                 # Strip out <nowiki> etc. added via replaceVariables
3795                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3796
3797                 # Signatures
3798                 $sigText = $this->getUserSig( $user );
3799                 $text = strtr( $text, array(
3800                         '~~~~~' => $d,
3801                         '~~~~' => "$sigText $d",
3802                         '~~~' => $sigText
3803                 ) );
3804
3805                 # Context links: [[|name]] and [[name (context)|]]
3806                 #
3807                 global $wgLegalTitleChars;
3808                 $tc = "[$wgLegalTitleChars]";
3809                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3810
3811                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3812                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3813                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3814
3815                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3816                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3817                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3818
3819                 $t = $this->mTitle->getText();
3820                 $m = array();
3821                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3822                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3823                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3824                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3825                 } else {
3826                         # if there's no context, don't bother duplicating the title
3827                         $text = preg_replace( $p2, '[[\\1]]', $text );
3828                 }
3829
3830                 # Trim trailing whitespace
3831                 $text = rtrim( $text );
3832
3833                 return $text;
3834         }
3835
3836         /**
3837          * Fetch the user's signature text, if any, and normalize to
3838          * validated, ready-to-insert wikitext.
3839          *
3840          * @param User $user
3841          * @return string
3842          * @private
3843          */
3844         function getUserSig( &$user ) {
3845                 global $wgMaxSigChars;
3846
3847                 $username = $user->getName();
3848                 $nickname = $user->getOption( 'nickname' );
3849                 $nickname = $nickname === '' ? $username : $nickname;
3850
3851                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3852                         $nickname = $username;
3853                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3854                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3855                         # Sig. might contain markup; validate this
3856                         if( $this->validateSig( $nickname ) !== false ) {
3857                                 # Validated; clean up (if needed) and return it
3858                                 return $this->cleanSig( $nickname, true );
3859                         } else {
3860                                 # Failed to validate; fall back to the default
3861                                 $nickname = $username;
3862                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3863                         }
3864                 }
3865
3866                 // Make sure nickname doesnt get a sig in a sig
3867                 $nickname = $this->cleanSigInSig( $nickname );
3868
3869                 # If we're still here, make it a link to the user page
3870                 $userpage = $user->getUserPage();
3871                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3872         }
3873
3874         /**
3875          * Check that the user's signature contains no bad XML
3876          *
3877          * @param string $text
3878          * @return mixed An expanded string, or false if invalid.
3879          */
3880         function validateSig( $text ) {
3881                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3882         }
3883
3884         /**
3885          * Clean up signature text
3886          *
3887          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3888          * 2) Substitute all transclusions
3889          *
3890          * @param string $text
3891          * @param $parsing Whether we're cleaning (preferences save) or parsing
3892          * @return string Signature text
3893          */
3894         function cleanSig( $text, $parsing = false ) {
3895                 global $wgTitle;
3896                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3897
3898                 $substWord = MagicWord::get( 'subst' );
3899                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3900                 $substText = '{{' . $substWord->getSynonym( 0 );
3901
3902                 $text = preg_replace( $substRegex, $substText, $text );
3903                 $text = $this->cleanSigInSig( $text );
3904                 $text = $this->replaceVariables( $text );
3905
3906                 $this->clearState();
3907                 return $text;
3908         }
3909
3910         /**
3911          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3912          * @param string $text
3913          * @return string Signature text with /~{3,5}/ removed
3914          */
3915         function cleanSigInSig( $text ) {
3916                 $text = preg_replace( '/~{3,5}/', '', $text );
3917                 return $text;
3918         }
3919
3920         /**
3921          * Set up some variables which are usually set up in parse()
3922          * so that an external function can call some class members with confidence
3923          * @public
3924          */
3925         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3926                 $this->mTitle =& $title;
3927                 $this->mOptions = $options;
3928                 $this->setOutputType( $outputType );
3929                 if ( $clearState ) {
3930                         $this->clearState();
3931                 }
3932         }
3933
3934         /**
3935          * Transform a MediaWiki message by replacing magic variables.
3936          *
3937          * @param string $text the text to transform
3938          * @param ParserOptions $options  options
3939          * @return string the text with variables substituted
3940          * @public
3941          */
3942         function transformMsg( $text, $options ) {
3943                 global $wgTitle;
3944                 static $executing = false;
3945
3946                 $fname = "Parser::transformMsg";
3947
3948                 # Guard against infinite recursion
3949                 if ( $executing ) {
3950                         return $text;
3951                 }
3952                 $executing = true;
3953
3954                 wfProfileIn($fname);
3955
3956                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3957                         $this->mTitle = $wgTitle;
3958                 } else {
3959                         $this->mTitle = Title::newFromText('msg');
3960                 }
3961                 $this->mOptions = $options;
3962                 $this->setOutputType( OT_MSG );
3963                 $this->clearState();
3964                 $text = $this->replaceVariables( $text );
3965
3966                 $executing = false;
3967                 wfProfileOut($fname);
3968                 return $text;
3969         }
3970
3971         /**
3972          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3973          * The callback should have the following form:
3974          *    function myParserHook( $text, $params, &$parser ) { ... }
3975          *
3976          * Transform and return $text. Use $parser for any required context, e.g. use
3977          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3978          *
3979          * @public
3980          *
3981          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3982          * @param mixed $callback The callback function (and object) to use for the tag
3983          *
3984          * @return The old value of the mTagHooks array associated with the hook
3985          */
3986         function setHook( $tag, $callback ) {
3987                 $tag = strtolower( $tag );
3988                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3989                 $this->mTagHooks[$tag] = $callback;
3990
3991                 return $oldVal;
3992         }
3993
3994         function setTransparentTagHook( $tag, $callback ) {
3995                 $tag = strtolower( $tag );
3996                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3997                 $this->mTransparentTagHooks[$tag] = $callback;
3998
3999                 return $oldVal;
4000         }
4001
4002         /**
4003          * Create a function, e.g. {{sum:1|2|3}}
4004          * The callback function should have the form:
4005          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4006          *
4007          * The callback may either return the text result of the function, or an array with the text
4008          * in element 0, and a number of flags in the other elements. The names of the flags are
4009          * specified in the keys. Valid flags are:
4010          *   found                     The text returned is valid, stop processing the template. This
4011          *                             is on by default.
4012          *   nowiki                    Wiki markup in the return value should be escaped
4013          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4014          *   noargs                    Don't replace triple-brace arguments in the return value
4015          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4016          *
4017          * @public
4018          *
4019          * @param string $id The magic word ID
4020          * @param mixed $callback The callback function (and object) to use
4021          * @param integer $flags a combination of the following flags:
4022          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4023          *
4024          * @return The old callback function for this name, if any
4025          */
4026         function setFunctionHook( $id, $callback, $flags = 0 ) {
4027                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
4028                 $this->mFunctionHooks[$id] = $callback;
4029
4030                 # Add to function cache
4031                 $mw = MagicWord::get( $id );
4032                 if( !$mw )
4033                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4034
4035                 $synonyms = $mw->getSynonyms();
4036                 $sensitive = intval( $mw->isCaseSensitive() );
4037
4038                 foreach ( $synonyms as $syn ) {
4039                         # Case
4040                         if ( !$sensitive ) {
4041                                 $syn = strtolower( $syn );
4042                         }
4043                         # Add leading hash
4044                         if ( !( $flags & SFH_NO_HASH ) ) {
4045                                 $syn = '#' . $syn;
4046                         }
4047                         # Remove trailing colon
4048                         if ( substr( $syn, -1, 1 ) == ':' ) {
4049                                 $syn = substr( $syn, 0, -1 );
4050                         }
4051                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4052                 }
4053                 return $oldVal;
4054         }
4055
4056         /**
4057          * Get all registered function hook identifiers
4058          *
4059          * @return array
4060          */
4061         function getFunctionHooks() {
4062                 return array_keys( $this->mFunctionHooks );
4063         }
4064
4065         /**
4066          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4067          * Placeholders created in Skin::makeLinkObj()
4068          * Returns an array of links found, indexed by PDBK:
4069          *  0 - broken
4070          *  1 - normal link
4071          *  2 - stub
4072          * $options is a bit field, RLH_FOR_UPDATE to select for update
4073          */
4074         function replaceLinkHolders( &$text, $options = 0 ) {
4075                 global $wgUser;
4076                 global $wgContLang;
4077
4078                 $fname = 'Parser::replaceLinkHolders';
4079                 wfProfileIn( $fname );
4080
4081                 $pdbks = array();
4082                 $colours = array();
4083                 $sk = $this->mOptions->getSkin();
4084                 $linkCache =& LinkCache::singleton();
4085
4086                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4087                         wfProfileIn( $fname.'-check' );
4088                         $dbr = wfGetDB( DB_SLAVE );
4089                         $page = $dbr->tableName( 'page' );
4090                         $threshold = $wgUser->getOption('stubthreshold');
4091
4092                         # Sort by namespace
4093                         asort( $this->mLinkHolders['namespaces'] );
4094
4095                         # Generate query
4096                         $query = false;
4097                         $current = null;
4098                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4099                                 # Make title object
4100                                 $title = $this->mLinkHolders['titles'][$key];
4101
4102                                 # Skip invalid entries.
4103                                 # Result will be ugly, but prevents crash.
4104                                 if ( is_null( $title ) ) {
4105                                         continue;
4106                                 }
4107                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4108
4109                                 # Check if it's a static known link, e.g. interwiki
4110                                 if ( $title->isAlwaysKnown() ) {
4111                                         $colours[$pdbk] = 1;
4112                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4113                                         $colours[$pdbk] = 1;
4114                                         $this->mOutput->addLink( $title, $id );
4115                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4116                                         $colours[$pdbk] = 0;
4117                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4118                                         $colours[$pdbk] = 0;
4119                                 } else {
4120                                         # Not in the link cache, add it to the query
4121                                         if ( !isset( $current ) ) {
4122                                                 $current = $ns;
4123                                                 $query =  "SELECT page_id, page_namespace, page_title";
4124                                                 if ( $threshold > 0 ) {
4125                                                         $query .= ', page_len, page_is_redirect';
4126                                                 }
4127                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4128                                         } elseif ( $current != $ns ) {
4129                                                 $current = $ns;
4130                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4131                                         } else {
4132                                                 $query .= ', ';
4133                                         }
4134
4135                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4136                                 }
4137                         }
4138                         if ( $query ) {
4139                                 $query .= '))';
4140                                 if ( $options & RLH_FOR_UPDATE ) {
4141                                         $query .= ' FOR UPDATE';
4142                                 }
4143
4144                                 $res = $dbr->query( $query, $fname );
4145
4146                                 # Fetch data and form into an associative array
4147                                 # non-existent = broken
4148                                 # 1 = known
4149                                 # 2 = stub
4150                                 while ( $s = $dbr->fetchObject($res) ) {
4151                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4152                                         $pdbk = $title->getPrefixedDBkey();
4153                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4154                                         $this->mOutput->addLink( $title, $s->page_id );
4155
4156                                         $colours[$pdbk] = ( $threshold == 0 || (
4157                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4158                                                                 $s->page_is_redirect ||
4159                                                                 !Namespace::isContent( $s->page_namespace ) )
4160                                                             ? 1 : 2 );
4161                                 }
4162                         }
4163                         wfProfileOut( $fname.'-check' );
4164
4165                         # Do a second query for different language variants of links and categories
4166                         if($wgContLang->hasVariants()){
4167                                 $linkBatch = new LinkBatch();
4168                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4169                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4170                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4171
4172                                 $categories = $this->mOutput->getCategoryLinks();
4173
4174                                 // Add variants of links to link batch
4175                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4176                                         $title = $this->mLinkHolders['titles'][$key];
4177                                         if ( is_null( $title ) )
4178                                                 continue;
4179
4180                                         $pdbk = $title->getPrefixedDBkey();
4181                                         $titleText = $title->getText();
4182
4183                                         // generate all variants of the link title text
4184                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4185
4186                                         // if link was not found (in first query), add all variants to query
4187                                         if ( !isset($colours[$pdbk]) ){
4188                                                 foreach($allTextVariants as $textVariant){
4189                                                         if($textVariant != $titleText){
4190                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4191                                                                 if(is_null($variantTitle)) continue;
4192                                                                 $linkBatch->addObj( $variantTitle );
4193                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4194                                                         }
4195                                                 }
4196                                         }
4197                                 }
4198
4199                                 // process categories, check if a category exists in some variant
4200                                 foreach( $categories as $category ){
4201                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4202                                         foreach($variants as $variant){
4203                                                 if($variant != $category){
4204                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4205                                                         if(is_null($variantTitle)) continue;
4206                                                         $linkBatch->addObj( $variantTitle );
4207                                                         $categoryMap[$variant] = $category;
4208                                                 }
4209                                         }
4210                                 }
4211
4212
4213                                 if(!$linkBatch->isEmpty()){
4214                                         // construct query
4215                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4216
4217                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4218                                         if ( $threshold > 0 ) {
4219                                                 $variantQuery .= ', page_len, page_is_redirect';
4220                                         }
4221
4222                                         $variantQuery .= " FROM $page WHERE $titleClause";
4223                                         if ( $options & RLH_FOR_UPDATE ) {
4224                                                 $variantQuery .= ' FOR UPDATE';
4225                                         }
4226
4227                                         $varRes = $dbr->query( $variantQuery, $fname );
4228
4229                                         // for each found variants, figure out link holders and replace
4230                                         while ( $s = $dbr->fetchObject($varRes) ) {
4231
4232                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4233                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4234                                                 $vardbk = $variantTitle->getDBkey();
4235
4236                                                 $holderKeys = array();
4237                                                 if(isset($variantMap[$varPdbk])){
4238                                                         $holderKeys = $variantMap[$varPdbk];
4239                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4240                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4241                                                 }
4242
4243                                                 // loop over link holders
4244                                                 foreach($holderKeys as $key){
4245                                                         $title = $this->mLinkHolders['titles'][$key];
4246                                                         if ( is_null( $title ) ) continue;
4247
4248                                                         $pdbk = $title->getPrefixedDBkey();
4249
4250                                                         if(!isset($colours[$pdbk])){
4251                                                                 // found link in some of the variants, replace the link holder data
4252                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4253                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4254
4255                                                                 // set pdbk and colour
4256                                                                 $pdbks[$key] = $varPdbk;
4257                                                                 if ( $threshold >  0 ) {
4258                                                                         $size = $s->page_len;
4259                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4260                                                                                 $colours[$varPdbk] = 1;
4261                                                                         } else {
4262                                                                                 $colours[$varPdbk] = 2;
4263                                                                         }
4264                                                                 }
4265                                                                 else {
4266                                                                         $colours[$varPdbk] = 1;
4267                                                                 }
4268                                                         }
4269                                                 }
4270
4271                                                 // check if the object is a variant of a category
4272                                                 if(isset($categoryMap[$vardbk])){
4273                                                         $oldkey = $categoryMap[$vardbk];
4274                                                         if($oldkey != $vardbk)
4275                                                                 $varCategories[$oldkey]=$vardbk;
4276                                                 }
4277                                         }
4278
4279                                         // rebuild the categories in original order (if there are replacements)
4280                                         if(count($varCategories)>0){
4281                                                 $newCats = array();
4282                                                 $originalCats = $this->mOutput->getCategories();
4283                                                 foreach($originalCats as $cat => $sortkey){
4284                                                         // make the replacement
4285                                                         if( array_key_exists($cat,$varCategories) )
4286                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4287                                                         else $newCats[$cat] = $sortkey;
4288                                                 }
4289                                                 $this->mOutput->setCategoryLinks($newCats);
4290                                         }
4291                                 }
4292                         }
4293
4294                         # Construct search and replace arrays
4295                         wfProfileIn( $fname.'-construct' );
4296                         $replacePairs = array();
4297                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4298                                 $pdbk = $pdbks[$key];
4299                                 $searchkey = "<!--LINK $key-->";
4300                                 $title = $this->mLinkHolders['titles'][$key];
4301                                 if ( empty( $colours[$pdbk] ) ) {
4302                                         $linkCache->addBadLinkObj( $title );
4303                                         $colours[$pdbk] = 0;
4304                                         $this->mOutput->addLink( $title, 0 );
4305                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4306                                                                         $this->mLinkHolders['texts'][$key],
4307                                                                         $this->mLinkHolders['queries'][$key] );
4308                                 } elseif ( $colours[$pdbk] == 1 ) {
4309                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4310                                                                         $this->mLinkHolders['texts'][$key],
4311                                                                         $this->mLinkHolders['queries'][$key] );
4312                                 } elseif ( $colours[$pdbk] == 2 ) {
4313                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4314                                                                         $this->mLinkHolders['texts'][$key],
4315                                                                         $this->mLinkHolders['queries'][$key] );
4316                                 }
4317                         }
4318                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4319                         wfProfileOut( $fname.'-construct' );
4320
4321                         # Do the thing
4322                         wfProfileIn( $fname.'-replace' );
4323                         $text = preg_replace_callback(
4324                                 '/(<!--LINK .*?-->)/',
4325                                 $replacer->cb(),
4326                                 $text);
4327
4328                         wfProfileOut( $fname.'-replace' );
4329                 }
4330
4331                 # Now process interwiki link holders
4332                 # This is quite a bit simpler than internal links
4333                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4334                         wfProfileIn( $fname.'-interwiki' );
4335                         # Make interwiki link HTML
4336                         $replacePairs = array();
4337                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4338                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4339                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4340                         }
4341                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4342
4343                         $text = preg_replace_callback(
4344                                 '/<!--IWLINK (.*?)-->/',
4345                                 $replacer->cb(),
4346                                 $text );
4347                         wfProfileOut( $fname.'-interwiki' );
4348                 }
4349
4350                 wfProfileOut( $fname );
4351                 return $colours;
4352         }
4353
4354         /**
4355          * Replace <!--LINK--> link placeholders with plain text of links
4356          * (not HTML-formatted).
4357          * @param string $text
4358          * @return string
4359          */
4360         function replaceLinkHoldersText( $text ) {
4361                 $fname = 'Parser::replaceLinkHoldersText';
4362                 wfProfileIn( $fname );
4363
4364                 $text = preg_replace_callback(
4365                         '/<!--(LINK|IWLINK) (.*?)-->/',
4366                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4367                         $text );
4368
4369                 wfProfileOut( $fname );
4370                 return $text;
4371         }
4372
4373         /**
4374          * @param array $matches
4375          * @return string
4376          * @private
4377          */
4378         function replaceLinkHoldersTextCallback( $matches ) {
4379                 $type = $matches[1];
4380                 $key  = $matches[2];
4381                 if( $type == 'LINK' ) {
4382                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4383                                 return $this->mLinkHolders['texts'][$key];
4384                         }
4385                 } elseif( $type == 'IWLINK' ) {
4386                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4387                                 return $this->mInterwikiLinkHolders['texts'][$key];
4388                         }
4389                 }
4390                 return $matches[0];
4391         }
4392
4393         /**
4394          * Tag hook handler for 'pre'.
4395          */
4396         function renderPreTag( $text, $attribs ) {
4397                 // Backwards-compatibility hack
4398                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4399
4400                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4401                 return wfOpenElement( 'pre', $attribs ) .
4402                         Xml::escapeTagsOnly( $content ) .
4403                         '</pre>';
4404         }
4405
4406         /**
4407          * Renders an image gallery from a text with one line per image.
4408          * text labels may be given by using |-style alternative text. E.g.
4409          *   Image:one.jpg|The number "1"
4410          *   Image:tree.jpg|A tree
4411          * given as text will return the HTML of a gallery with two images,
4412          * labeled 'The number "1"' and
4413          * 'A tree'.
4414          */
4415         function renderImageGallery( $text, $params ) {
4416                 $ig = new ImageGallery();
4417                 $ig->setContextTitle( $this->mTitle );
4418                 $ig->setShowBytes( false );
4419                 $ig->setShowFilename( false );
4420                 $ig->setParsing();
4421                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4422                 $ig->useSkin( $this->mOptions->getSkin() );
4423                 $ig->mRevisionId = $this->mRevisionId;
4424
4425                 if( isset( $params['caption'] ) ) {
4426                         $caption = $params['caption'];
4427                         $caption = htmlspecialchars( $caption );
4428                         $caption = $this->replaceInternalLinks( $caption );
4429                         $ig->setCaptionHtml( $caption );
4430                 }
4431                 if( isset( $params['perrow'] ) ) {
4432                         $ig->setPerRow( $params['perrow'] );
4433                 }
4434                 if( isset( $params['widths'] ) ) {
4435                         $ig->setWidths( $params['widths'] );
4436                 }
4437                 if( isset( $params['heights'] ) ) {
4438                         $ig->setHeights( $params['heights'] );
4439                 }
4440
4441                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4442
4443                 $lines = explode( "\n", $text );
4444                 foreach ( $lines as $line ) {
4445                         # match lines like these:
4446                         # Image:someimage.jpg|This is some image
4447                         $matches = array();
4448                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4449                         # Skip empty lines
4450                         if ( count( $matches ) == 0 ) {
4451                                 continue;
4452                         }
4453                         $tp = Title::newFromText( $matches[1] );
4454                         $nt =& $tp;
4455                         if( is_null( $nt ) ) {
4456                                 # Bogus title. Ignore these so we don't bomb out later.
4457                                 continue;
4458                         }
4459                         if ( isset( $matches[3] ) ) {
4460                                 $label = $matches[3];
4461                         } else {
4462                                 $label = '';
4463                         }
4464
4465                         $pout = $this->parse( $label,
4466                                 $this->mTitle,
4467                                 $this->mOptions,
4468                                 false, // Strip whitespace...?
4469                                 false  // Don't clear state!
4470                         );
4471                         $html = $pout->getText();
4472
4473                         $ig->add( $nt, $html );
4474
4475                         # Only add real images (bug #5586)
4476                         if ( $nt->getNamespace() == NS_IMAGE ) {
4477                                 $this->mOutput->addImage( $nt->getDBkey() );
4478                         }
4479                 }
4480                 return $ig->toHTML();
4481         }
4482
4483         /**
4484          * Parse image options text and use it to make an image
4485          */
4486         function makeImage( $nt, $options ) {
4487                 # @TODO: let the MediaHandler specify its transform parameters
4488                 #
4489                 # Check if the options text is of the form "options|alt text"
4490                 # Options are:
4491                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4492                 #  * left               no resizing, just left align. label is used for alt= only
4493                 #  * right              same, but right aligned
4494                 #  * none               same, but not aligned
4495                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4496                 #  * center             center the image
4497                 #  * framed             Keep original image size, no magnify-button.
4498                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4499                 #  * upright            reduce width for upright images, rounded to full __0 px
4500                 #  * border             draw a 1px border around the image
4501                 # vertical-align values (no % or length right now):
4502                 #  * baseline
4503                 #  * sub
4504                 #  * super
4505                 #  * top
4506                 #  * text-top
4507                 #  * middle
4508                 #  * bottom
4509                 #  * text-bottom
4510
4511
4512                 $part = array_map( 'trim', explode( '|', $options) );
4513
4514                 $mwAlign = array();
4515                 $alignments = array( 'left', 'right', 'center', 'none', 'baseline', 'sub', 'super', 'top', 'text-top', 'middle', 'bottom', 'text-bottom' );
4516                 foreach ( $alignments as $alignment ) {
4517                         $mwAlign[$alignment] =& MagicWord::get( 'img_'.$alignment );
4518                 }
4519                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4520                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4521                 $mwWidth  =& MagicWord::get( 'img_width' );
4522                 $mwFramed =& MagicWord::get( 'img_framed' );
4523                 $mwFrameless =& MagicWord::get( 'img_frameless' );
4524                 $mwUpright =& MagicWord::get( 'img_upright' );
4525                 $mwBorder =& MagicWord::get( 'img_border' );
4526                 $mwPage   =& MagicWord::get( 'img_page' );
4527                 $caption = '';
4528
4529                 $params = array();
4530                 $framed = $thumb = false;
4531                 $manual_thumb = '' ;
4532                 $align = $valign = '';
4533                 $sk = $this->mOptions->getSkin();
4534
4535                 foreach( $part as $val ) {
4536                         if ( !is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4537                                 $thumb=true;
4538                         } elseif ( !is_null( $match = $mwUpright->matchVariableStartToEnd( $val ) ) ) {
4539                                 $params['upright'] = true;
4540                                 $params['upright_factor'] = floatval( $match );
4541                         } elseif ( !is_null( $match = $mwFrameless->matchVariableStartToEnd( $val ) ) ) {
4542                                 $params['frameless'] = true;
4543                         } elseif ( !is_null( $mwBorder->matchVariableStartToEnd( $val ) ) ) {
4544                                 $params['border'] = true;
4545                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4546                                 # use manually specified thumbnail
4547                                 $thumb=true;
4548                                 $manual_thumb = $match;
4549                         } else {
4550                                 foreach( $alignments as $alignment ) {
4551                                         if ( ! is_null( $mwAlign[$alignment]->matchVariableStartToEnd($val) ) ) {
4552                                                 switch ( $alignment ) {
4553                                                         case 'left': case 'right': case 'center': case 'none':
4554                                                                 $align = $alignment; break;
4555                                                         default:
4556                                                                 $valign = $alignment;
4557                                                 }
4558                                                 continue 2;
4559                                         }
4560                                 }
4561                                 if ( ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4562                                         # Select a page in a multipage document
4563                                         $params['page'] = $match;
4564                                 } elseif ( !isset( $params['width'] ) && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4565                                         wfDebug( "img_width match: $match\n" );
4566                                         # $match is the image width in pixels
4567                                         $m = array();
4568                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4569                                                  $params['width'] = intval( $m[1] );
4570                                                  $params['height'] = intval( $m[2] );
4571                                         } else {
4572                                                 $params['width'] = intval($match);
4573                                         }
4574                                 } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4575                                         $framed=true;
4576                                 } else {
4577                                         $caption = $val;
4578                                 }
4579                         }
4580                 }
4581                 # Strip bad stuff out of the alt text
4582                 $alt = $this->replaceLinkHoldersText( $caption );
4583
4584                 # make sure there are no placeholders in thumbnail attributes
4585                 # that are later expanded to html- so expand them now and
4586                 # remove the tags
4587                 $alt = $this->mStripState->unstripBoth( $alt );
4588                 $alt = Sanitizer::stripAllTags( $alt );
4589
4590                 # Give extensions a chance to select the file revision for us
4591                 $skip = $time = false;
4592                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
4593
4594                 # Linker does the rest
4595                 if( $skip ) {
4596                         $link = $sk->makeLinkObj( $nt );
4597                 } else {
4598                         $link = $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $params, $framed, $thumb, $manual_thumb, $valign, $time );
4599                 }
4600
4601                 return $link;
4602         }
4603
4604         /**
4605          * Set a flag in the output object indicating that the content is dynamic and
4606          * shouldn't be cached.
4607          */
4608         function disableCache() {
4609                 wfDebug( "Parser output marked as uncacheable.\n" );
4610                 $this->mOutput->mCacheTime = -1;
4611         }
4612
4613         /**#@+
4614          * Callback from the Sanitizer for expanding items found in HTML attribute
4615          * values, so they can be safely tested and escaped.
4616          * @param string $text
4617          * @param array $args
4618          * @return string
4619          * @private
4620          */
4621         function attributeStripCallback( &$text, $args ) {
4622                 $text = $this->replaceVariables( $text, $args );
4623                 $text = $this->mStripState->unstripBoth( $text );
4624                 return $text;
4625         }
4626
4627         /**#@-*/
4628
4629         /**#@+
4630          * Accessor/mutator
4631          */
4632         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4633         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4634         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4635         /**#@-*/
4636
4637         /**#@+
4638          * Accessor
4639          */
4640         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4641         /**#@-*/
4642
4643
4644         /**
4645          * Break wikitext input into sections, and either pull or replace
4646          * some particular section's text.
4647          *
4648          * External callers should use the getSection and replaceSection methods.
4649          *
4650          * @param $text Page wikitext
4651          * @param $section Numbered section. 0 pulls the text before the first
4652          *                 heading; other numbers will pull the given section
4653          *                 along with its lower-level subsections.
4654          * @param $mode One of "get" or "replace"
4655          * @param $newtext Replacement text for section data.
4656          * @return string for "get", the extracted section text.
4657          *                for "replace", the whole page with the section replaced.
4658          */
4659         private function extractSections( $text, $section, $mode, $newtext='' ) {
4660                 # I.... _hope_ this is right.
4661                 # Otherwise, sometimes we don't have things initialized properly.
4662                 $this->clearState();
4663
4664                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4665                 # comments to be stripped as well)
4666                 $stripState = new StripState;
4667
4668                 $oldOutputType = $this->mOutputType;
4669                 $oldOptions = $this->mOptions;
4670                 $this->mOptions = new ParserOptions();
4671                 $this->setOutputType( OT_WIKI );
4672
4673                 $striptext = $this->strip( $text, $stripState, true );
4674
4675                 $this->setOutputType( $oldOutputType );
4676                 $this->mOptions = $oldOptions;
4677
4678                 # now that we can be sure that no pseudo-sections are in the source,
4679                 # split it up by section
4680                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4681                 $comment = "(?:$uniq-!--.*?QINU\x07)";
4682                 $secs = preg_split(
4683                         "/
4684                         (
4685                                 ^
4686                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4687                                 (=+) # Should this be limited to 6?
4688                                 .+?  # Section title...
4689                                 \\2  # Ending = count must match start
4690                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4691                                 $
4692                         |
4693                                 <h([1-6])\b.*?>
4694                                 .*?
4695                                 <\/h\\3\s*>
4696                         )
4697                         /mix",
4698                         $striptext, -1,
4699                         PREG_SPLIT_DELIM_CAPTURE);
4700
4701                 if( $mode == "get" ) {
4702                         if( $section == 0 ) {
4703                                 // "Section 0" returns the content before any other section.
4704                                 $rv = $secs[0];
4705                         } else {
4706                                 //track missing section, will replace if found.
4707                                 $rv = $newtext;
4708                         }
4709                 } elseif( $mode == "replace" ) {
4710                         if( $section == 0 ) {
4711                                 $rv = $newtext . "\n\n";
4712                                 $remainder = true;
4713                         } else {
4714                                 $rv = $secs[0];
4715                                 $remainder = false;
4716                         }
4717                 }
4718                 $count = 0;
4719                 $sectionLevel = 0;
4720                 for( $index = 1; $index < count( $secs ); ) {
4721                         $headerLine = $secs[$index++];
4722                         if( $secs[$index] ) {
4723                                 // A wiki header
4724                                 $headerLevel = strlen( $secs[$index++] );
4725                         } else {
4726                                 // An HTML header
4727                                 $index++;
4728                                 $headerLevel = intval( $secs[$index++] );
4729                         }
4730                         $content = $secs[$index++];
4731
4732                         $count++;
4733                         if( $mode == "get" ) {
4734                                 if( $count == $section ) {
4735                                         $rv = $headerLine . $content;
4736                                         $sectionLevel = $headerLevel;
4737                                 } elseif( $count > $section ) {
4738                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4739                                                 $rv .= $headerLine . $content;
4740                                         } else {
4741                                                 // Broke out to a higher-level section
4742                                                 break;
4743                                         }
4744                                 }
4745                         } elseif( $mode == "replace" ) {
4746                                 if( $count < $section ) {
4747                                         $rv .= $headerLine . $content;
4748                                 } elseif( $count == $section ) {
4749                                         $rv .= $newtext . "\n\n";
4750                                         $sectionLevel = $headerLevel;
4751                                 } elseif( $count > $section ) {
4752                                         if( $headerLevel <= $sectionLevel ) {
4753                                                 // Passed the section's sub-parts.
4754                                                 $remainder = true;
4755                                         }
4756                                         if( $remainder ) {
4757                                                 $rv .= $headerLine . $content;
4758                                         }
4759                                 }
4760                         }
4761                 }
4762                 if (is_string($rv))
4763                         # reinsert stripped tags
4764                         $rv = trim( $stripState->unstripBoth( $rv ) );
4765
4766                 return $rv;
4767         }
4768
4769         /**
4770          * This function returns the text of a section, specified by a number ($section).
4771          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4772          * the first section before any such heading (section 0).
4773          *
4774          * If a section contains subsections, these are also returned.
4775          *
4776          * @param $text String: text to look in
4777          * @param $section Integer: section number
4778          * @param $deftext: default to return if section is not found
4779          * @return string text of the requested section
4780          */
4781         public function getSection( $text, $section, $deftext='' ) {
4782                 return $this->extractSections( $text, $section, "get", $deftext );
4783         }
4784
4785         public function replaceSection( $oldtext, $section, $text ) {
4786                 return $this->extractSections( $oldtext, $section, "replace", $text );
4787         }
4788
4789         /**
4790          * Get the timestamp associated with the current revision, adjusted for
4791          * the default server-local timestamp
4792          */
4793         function getRevisionTimestamp() {
4794                 if ( is_null( $this->mRevisionTimestamp ) ) {
4795                         wfProfileIn( __METHOD__ );
4796                         global $wgContLang;
4797                         $dbr = wfGetDB( DB_SLAVE );
4798                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4799                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4800
4801                         // Normalize timestamp to internal MW format for timezone processing.
4802                         // This has the added side-effect of replacing a null value with
4803                         // the current time, which gives us more sensible behavior for
4804                         // previews.
4805                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4806
4807                         // The cryptic '' timezone parameter tells to use the site-default
4808                         // timezone offset instead of the user settings.
4809                         //
4810                         // Since this value will be saved into the parser cache, served
4811                         // to other users, and potentially even used inside links and such,
4812                         // it needs to be consistent for all visitors.
4813                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4814
4815                         wfProfileOut( __METHOD__ );
4816                 }
4817                 return $this->mRevisionTimestamp;
4818         }
4819
4820         /**
4821          * Mutator for $mDefaultSort
4822          *
4823          * @param $sort New value
4824          */
4825         public function setDefaultSort( $sort ) {
4826                 $this->mDefaultSort = $sort;
4827         }
4828
4829         /**
4830          * Accessor for $mDefaultSort
4831          * Will use the title/prefixed title if none is set
4832          *
4833          * @return string
4834          */
4835         public function getDefaultSort() {
4836                 if( $this->mDefaultSort !== false ) {
4837                         return $this->mDefaultSort;
4838                 } else {
4839                         return $this->mTitle->getNamespace() == NS_CATEGORY
4840                                         ? $this->mTitle->getText()
4841                                         : $this->mTitle->getPrefixedText();
4842                 }
4843         }
4844
4845 }
4846
4847 /**
4848  * @todo document, briefly.
4849  * @addtogroup Parser
4850  */
4851 class OnlyIncludeReplacer {
4852         var $output = '';
4853
4854         function replace( $matches ) {
4855                 if ( substr( $matches[1], -1 ) == "\n" ) {
4856                         $this->output .= substr( $matches[1], 0, -1 );
4857                 } else {
4858                         $this->output .= $matches[1];
4859                 }
4860         }
4861 }
4862
4863 /**
4864  * @todo document, briefly.
4865  * @addtogroup Parser
4866  */
4867 class StripState {
4868         var $general, $nowiki;
4869
4870         function __construct() {
4871                 $this->general = new ReplacementArray;
4872                 $this->nowiki = new ReplacementArray;
4873         }
4874
4875         function unstripGeneral( $text ) {
4876                 wfProfileIn( __METHOD__ );
4877                 $text = $this->general->replace( $text );
4878                 wfProfileOut( __METHOD__ );
4879                 return $text;
4880         }
4881
4882         function unstripNoWiki( $text ) {
4883                 wfProfileIn( __METHOD__ );
4884                 $text = $this->nowiki->replace( $text );
4885                 wfProfileOut( __METHOD__ );
4886                 return $text;
4887         }
4888
4889         function unstripBoth( $text ) {
4890                 wfProfileIn( __METHOD__ );
4891                 $text = $this->general->replace( $text );
4892                 $text = $this->nowiki->replace( $text );
4893                 wfProfileOut( __METHOD__ );
4894                 return $text;
4895         }
4896 }
4897
4898 ?>