includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are four main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * transformMsg()
  24  *   performs brace substitution on MediaWiki messages
  25  * preprocess()
  26  *   removes HTML comments and expands templates
  27  *
  28  * Globals used:
  29  *    objects:   $wgLang, $wgContLang
  30  *
  31  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  32  *
  33  * settings:
  34  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  35  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  36  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  37  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.2';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  62                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  63
  64         // State constants for the definition list colon extraction
  65         const COLON_STATE_TEXT = 0;
  66         const COLON_STATE_TAG = 1;
  67         const COLON_STATE_TAGSTART = 2;
  68         const COLON_STATE_CLOSETAG = 3;
  69         const COLON_STATE_TAGSLASH = 4;
  70         const COLON_STATE_COMMENT = 5;
  71         const COLON_STATE_COMMENTDASH = 6;
  72         const COLON_STATE_COMMENTDASHDASH = 7;
  73
  74         /**#@+
  75          * @private
  76          */
  77         # Persistent:
  78         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  79                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  80                 $mExtLinkBracketedRegex;
  81
  82         # Cleared with clearState():
  83         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  84         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  85         var $mInterwikiLinkHolders, $mLinkHolders;
  86         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  87         var $mTplExpandCache,// empty-frame expansion cache
  88             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  89                                 // in this path. Used for loop detection.
  90         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  91
  92         # Temporary
  93         # These are variables reset at least once per parse regardless of $clearState
  94         var $mOptions,      // ParserOptions object
  95                 $mTitle,        // Title context, used for self-link rendering and similar things
  96                 $mOutputType,   // Output type, one of the OT_xxx constants
  97                 $ot,            // Shortcut alias, see setOutputType()
  98                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  99                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 100                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 101
 102         /**#@-*/
 103
 104         /**
 105          * Constructor
 106          *
 107          * @public
 108          */
 109         function __construct( $conf = array() ) {
 110                 $this->mTagHooks = array();
 111                 $this->mTransparentTagHooks = array();
 112                 $this->mFunctionHooks = array();
 113                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 114                 $this->mStripList = array( 'nowiki', 'gallery' );
 115                 $this->mMarkerSuffix = "-QINU\x7f";
 116                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 117                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 118                 $this->mFirstCall = true;
 119         }
 120
 121         /**
 122          * Do various kinds of initialisation on the first call of the parser
 123          */
 124         function firstCallInit() {
 125                 if ( !$this->mFirstCall ) {
 126                         return;
 127                 }
 128
 129                 wfProfileIn( __METHOD__ );
 130                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 131
 132                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 133
 134                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 135                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 136                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 137                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 138                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 139                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 140                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 141                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 142                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 159                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'filepath', array( 'CoreParserFunctions', 'filepath' ), SFH_NO_HASH );
 161
 162                 if ( $wgAllowDisplayTitle ) {
 163                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 164                 }
 165                 if ( $wgAllowSlowParserFunctions ) {
 166                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 167                 }
 168
 169                 $this->initialiseVariables();
 170                 $this->mFirstCall = false;
 171                 wfProfileOut( __METHOD__ );
 172         }
 173
 174         /**
 175          * Clear Parser state
 176          *
 177          * @private
 178          */
 179         function clearState() {
 180                 wfProfileIn( __METHOD__ );
 181                 if ( $this->mFirstCall ) {
 182                         $this->firstCallInit();
 183                 }
 184                 $this->mOutput = new ParserOutput;
 185                 $this->mAutonumber = 0;
 186                 $this->mLastSection = '';
 187                 $this->mDTopen = false;
 188                 $this->mIncludeCount = array();
 189                 $this->mStripState = new StripState;
 190                 $this->mArgStack = false;
 191                 $this->mInPre = false;
 192                 $this->mInterwikiLinkHolders = array(
 193                         'texts' => array(),
 194                         'titles' => array()
 195                 );
 196                 $this->mLinkHolders = array(
 197                         'namespaces' => array(),
 198                         'dbkeys' => array(),
 199                         'queries' => array(),
 200                         'texts' => array(),
 201                         'titles' => array()
 202                 );
 203                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 204
 205                 /**
 206                  * Prefix for temporary replacement strings for the multipass parser.
 207                  * \x07 should never appear in input as it's disallowed in XML.
 208                  * Using it at the front also gives us a little extra robustness
 209                  * since it shouldn't match when butted up against identifier-like
 210                  * string constructs.
 211                  *
 212                  * Must not consist of all title characters, or else it will change
 213                  * the behaviour of <nowiki> in a link.
 214                  */
 215                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 216                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTemplatePath = array();
 220                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 221
 222                 $this->mShowToc = true;
 223                 $this->mForceTocPosition = false;
 224                 $this->mIncludeSizes = array(
 225                         'post-expand' => 0,
 226                         'arg' => 0,
 227                 );
 228                 $this->mPPNodeCount = 0;
 229                 $this->mDefaultSort = false;
 230                 $this->mHeadings = array();
 231
 232                 wfRunHooks( 'ParserClearState', array( &$this ) );
 233                 wfProfileOut( __METHOD__ );
 234         }
 235
 236         function setOutputType( $ot ) {
 237                 $this->mOutputType = $ot;
 238                 // Shortcut alias
 239                 $this->ot = array(
 240                         'html' => $ot == OT_HTML,
 241                         'wiki' => $ot == OT_WIKI,
 242                         'msg' => $ot == OT_MSG,
 243                         'pre' => $ot == OT_PREPROCESS,
 244                 );
 245         }
 246
 247         /**
 248          * Accessor for mUniqPrefix.
 249          *
 250          * @public
 251          */
 252         function uniqPrefix() {
 253                 return $this->mUniqPrefix;
 254         }
 255
 256         /**
 257          * Convert wikitext to HTML
 258          * Do not call this function recursively.
 259          *
 260          * @param string $text Text we want to parse
 261          * @param Title &$title A title object
 262          * @param array $options
 263          * @param boolean $linestart
 264          * @param boolean $clearState
 265          * @param int $revid number to pass in {{REVISIONID}}
 266          * @return ParserOutput a ParserOutput
 267          */
 268         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 269                 /**
 270                  * First pass--just handle <nowiki> sections, pass the rest off
 271                  * to internalParse() which does all the real work.
 272                  */
 273
 274                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 275                 $fname = 'Parser::parse-' . wfGetCaller();
 276                 wfProfileIn( __METHOD__ );
 277                 wfProfileIn( $fname );
 278
 279                 if ( $clearState ) {
 280                         $this->clearState();
 281                 }
 282
 283                 $this->mOptions = $options;
 284                 $this->mTitle =& $title;
 285                 $oldRevisionId = $this->mRevisionId;
 286                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 287                 if( $revid !== null ) {
 288                         $this->mRevisionId = $revid;
 289                         $this->mRevisionTimestamp = null;
 290                 }
 291                 $this->setOutputType( OT_HTML );
 292                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 293                 # No more strip!
 294                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 295                 $text = $this->internalParse( $text );
 296                 $text = $this->mStripState->unstripGeneral( $text );
 297
 298                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 299                 $fixtags = array(
 300                         # french spaces, last one Guillemet-left
 301                         # only if there is something before the space
 302                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 303                         # french spaces, Guillemet-right
 304                         '/(\\302\\253) /' => '\\1&nbsp;',
 305                 );
 306                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 307
 308                 # only once and last
 309                 $text = $this->doBlockLevels( $text, $linestart );
 310
 311                 $this->replaceLinkHolders( $text );
 312
 313                 # the position of the parserConvert() call should not be changed. it
 314                 # assumes that the links are all replaced and the only thing left
 315                 # is the <nowiki> mark.
 316                 # Side-effects: this calls $this->mOutput->setTitleText()
 317                 $text = $wgContLang->parserConvert( $text, $this );
 318
 319                 $text = $this->mStripState->unstripNoWiki( $text );
 320
 321                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 322
 323 //!JF Move to its own function
 324
 325                 $uniq_prefix = $this->mUniqPrefix;
 326                 $matches = array();
 327                 $elements = array_keys( $this->mTransparentTagHooks );
 328                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 329
 330                 foreach( $matches as $marker => $data ) {
 331                         list( $element, $content, $params, $tag ) = $data;
 332                         $tagName = strtolower( $element );
 333                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 334                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 335                                         array( $content, $params, $this ) );
 336                         } else {
 337                                 $output = $tag;
 338                         }
 339                         $this->mStripState->general->setPair( $marker, $output );
 340                 }
 341                 $text = $this->mStripState->unstripGeneral( $text );
 342
 343                 $text = Sanitizer::normalizeCharReferences( $text );
 344
 345                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 346                         $text = Parser::tidy($text);
 347                 } else {
 348                         # attempt to sanitize at least some nesting problems
 349                         # (bug #2702 and quite a few others)
 350                         $tidyregs = array(
 351                                 # ''Something [http://www.cool.com cool''] -->
 352                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 353                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 354                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 355                                 # fix up an anchor inside another anchor, only
 356                                 # at least for a single single nested link (bug 3695)
 357                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 358                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 359                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 360                                 # contains a div, so fix it up here; replace
 361                                 # div with escaped text
 362                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 363                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 364                                 # remove empty italic or bold tag pairs, some
 365                                 # introduced by rules above
 366                                 '/<([bi])><\/\\1>/' => '',
 367                         );
 368
 369                         $text = preg_replace(
 370                                 array_keys( $tidyregs ),
 371                                 array_values( $tidyregs ),
 372                                 $text );
 373                 }
 374
 375                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 376
 377                 # Information on include size limits, for the benefit of users who try to skirt them
 378                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 379                         $max = $this->mOptions->getMaxIncludeSize();
 380                         $text .= "<!-- \n" .
 381                                 "Preprocessor node count: {$this->mPPNodeCount}\n" .
 382                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 383                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 384                                 "Maximum: $max bytes\n" .
 385                                 "-->\n";
 386                 }
 387                 $this->mOutput->setText( $text );
 388                 $this->mRevisionId = $oldRevisionId;
 389                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 390                 wfProfileOut( $fname );
 391                 wfProfileOut( __METHOD__ );
 392
 393                 return $this->mOutput;
 394         }
 395
 396         /**
 397          * Recursive parser entry point that can be called from an extension tag
 398          * hook.
 399          */
 400         function recursiveTagParse( $text ) {
 401                 wfProfileIn( __METHOD__ );
 402                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 403                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 404                 $text = $this->internalParse( $text );
 405                 wfProfileOut( __METHOD__ );
 406                 return $text;
 407         }
 408
 409         /**
 410          * Expand templates and variables in the text, producing valid, static wikitext.
 411          * Also removes comments.
 412          */
 413         function preprocess( $text, $title, $options, $revid = null ) {
 414                 wfProfileIn( __METHOD__ );
 415                 $this->clearState();
 416                 $this->setOutputType( OT_PREPROCESS );
 417                 $this->mOptions = $options;
 418                 $this->mTitle = $title;
 419                 if( $revid !== null ) {
 420                         $this->mRevisionId = $revid;
 421                 }
 422                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 423                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 424                 $text = $this->replaceVariables( $text );
 425                 if ( $this->mOptions->getRemoveComments() ) {
 426                         $text = Sanitizer::removeHTMLcomments( $text );
 427                 }
 428                 $text = $this->mStripState->unstripBoth( $text );
 429                 wfProfileOut( __METHOD__ );
 430                 return $text;
 431         }
 432
 433         /**
 434          * Get a random string
 435          *
 436          * @private
 437          * @static
 438          */
 439         function getRandomString() {
 440                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 441         }
 442
 443         function &getTitle() { return $this->mTitle; }
 444         function getOptions() { return $this->mOptions; }
 445
 446         function getFunctionLang() {
 447                 global $wgLang, $wgContLang;
 448                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 449         }
 450
 451         /**
 452          * Replaces all occurrences of HTML-style comments and the given tags
 453          * in the text with a random marker and returns teh next text. The output
 454          * parameter $matches will be an associative array filled with data in
 455          * the form:
 456          *   'UNIQ-xxxxx' => array(
 457          *     'element',
 458          *     'tag content',
 459          *     array( 'param' => 'x' ),
 460          *     '<element param="x">tag content</element>' ) )
 461          *
 462          * @param $elements list of element names. Comments are always extracted.
 463          * @param $text Source text string.
 464          * @param $uniq_prefix
 465          *
 466          * @public
 467          * @static
 468          */
 469         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 470                 static $n = 1;
 471                 $stripped = '';
 472                 $matches = array();
 473
 474                 $taglist = implode( '|', $elements );
 475                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 476
 477                 while ( '' != $text ) {
 478                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 479                         $stripped .= $p[0];
 480                         if( count( $p ) < 5 ) {
 481                                 break;
 482                         }
 483                         if( count( $p ) > 5 ) {
 484                                 // comment
 485                                 $element    = $p[4];
 486                                 $attributes = '';
 487                                 $close      = '';
 488                                 $inside     = $p[5];
 489                         } else {
 490                                 // tag
 491                                 $element    = $p[1];
 492                                 $attributes = $p[2];
 493                                 $close      = $p[3];
 494                                 $inside     = $p[4];
 495                         }
 496
 497                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 498                         $stripped .= $marker;
 499
 500                         if ( $close === '/>' ) {
 501                                 // Empty element tag, <tag />
 502                                 $content = null;
 503                                 $text = $inside;
 504                                 $tail = null;
 505                         } else {
 506                                 if( $element == '!--' ) {
 507                                         $end = '/(-->)/';
 508                                 } else {
 509                                         $end = "/(<\\/$element\\s*>)/i";
 510                                 }
 511                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 512                                 $content = $q[0];
 513                                 if( count( $q ) < 3 ) {
 514                                         # No end tag -- let it run out to the end of the text.
 515                                         $tail = '';
 516                                         $text = '';
 517                                 } else {
 518                                         $tail = $q[1];
 519                                         $text = $q[2];
 520                                 }
 521                         }
 522
 523                         $matches[$marker] = array( $element,
 524                                 $content,
 525                                 Sanitizer::decodeTagAttributes( $attributes ),
 526                                 "<$element$attributes$close$content$tail" );
 527                 }
 528                 return $stripped;
 529         }
 530
 531         /**
 532          * Get a list of strippable XML-like elements
 533          */
 534         function getStripList() {
 535                 global $wgRawHtml;
 536                 $elements = $this->mStripList;
 537                 if( $wgRawHtml ) {
 538                         $elements[] = 'html';
 539                 }
 540                 if( $this->mOptions->getUseTeX() ) {
 541                         $elements[] = 'math';
 542                 }
 543                 return $elements;
 544         }
 545
 546         /**
 547          * @deprecated use replaceVariables
 548          */
 549         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 550                 return $text;
 551         }
 552
 553         /**
 554          * Restores pre, math, and other extensions removed by strip()
 555          *
 556          * always call unstripNoWiki() after this one
 557          * @private
 558          * @deprecated use $this->mStripState->unstrip()
 559          */
 560         function unstrip( $text, $state ) {
 561                 return $state->unstripGeneral( $text );
 562         }
 563
 564         /**
 565          * Always call this after unstrip() to preserve the order
 566          *
 567          * @private
 568          * @deprecated use $this->mStripState->unstrip()
 569          */
 570         function unstripNoWiki( $text, $state ) {
 571                 return $state->unstripNoWiki( $text );
 572         }
 573
 574         /**
 575          * @deprecated use $this->mStripState->unstripBoth()
 576          */
 577         function unstripForHTML( $text ) {
 578                 return $this->mStripState->unstripBoth( $text );
 579         }
 580
 581         /**
 582          * Add an item to the strip state
 583          * Returns the unique tag which must be inserted into the stripped text
 584          * The tag will be replaced with the original text in unstrip()
 585          *
 586          * @private
 587          */
 588         function insertStripItem( $text ) {
 589                 static $n = 0;
 590                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 591                 $this->mStripState->general->setPair( $rnd, $text );
 592                 return $rnd;
 593         }
 594
 595         /**
 596          * Interface with html tidy, used if $wgUseTidy = true.
 597          * If tidy isn't able to correct the markup, the original will be
 598          * returned in all its glory with a warning comment appended.
 599          *
 600          * Either the external tidy program or the in-process tidy extension
 601          * will be used depending on availability. Override the default
 602          * $wgTidyInternal setting to disable the internal if it's not working.
 603          *
 604          * @param string $text Hideous HTML input
 605          * @return string Corrected HTML output
 606          * @public
 607          * @static
 608          */
 609         function tidy( $text ) {
 610                 global $wgTidyInternal;
 611                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 612 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 613 '<head><title>test</title></head><body>'.$text.'</body></html>';
 614                 if( $wgTidyInternal ) {
 615                         $correctedtext = Parser::internalTidy( $wrappedtext );
 616                 } else {
 617                         $correctedtext = Parser::externalTidy( $wrappedtext );
 618                 }
 619                 if( is_null( $correctedtext ) ) {
 620                         wfDebug( "Tidy error detected!\n" );
 621                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 622                 }
 623                 return $correctedtext;
 624         }
 625
 626         /**
 627          * Spawn an external HTML tidy process and get corrected markup back from it.
 628          *
 629          * @private
 630          * @static
 631          */
 632         function externalTidy( $text ) {
 633                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 634                 $fname = 'Parser::externalTidy';
 635                 wfProfileIn( $fname );
 636
 637                 $cleansource = '';
 638                 $opts = ' -utf8';
 639
 640                 $descriptorspec = array(
 641                         0 => array('pipe', 'r'),
 642                         1 => array('pipe', 'w'),
 643                         2 => array('file', wfGetNull(), 'a')
 644                 );
 645                 $pipes = array();
 646                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 647                 if (is_resource($process)) {
 648                         // Theoretically, this style of communication could cause a deadlock
 649                         // here. If the stdout buffer fills up, then writes to stdin could
 650                         // block. This doesn't appear to happen with tidy, because tidy only
 651                         // writes to stdout after it's finished reading from stdin. Search
 652                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 653                         fwrite($pipes[0], $text);
 654                         fclose($pipes[0]);
 655                         while (!feof($pipes[1])) {
 656                                 $cleansource .= fgets($pipes[1], 1024);
 657                         }
 658                         fclose($pipes[1]);
 659                         proc_close($process);
 660                 }
 661
 662                 wfProfileOut( $fname );
 663
 664                 if( $cleansource == '' && $text != '') {
 665                         // Some kind of error happened, so we couldn't get the corrected text.
 666                         // Just give up; we'll use the source text and append a warning.
 667                         return null;
 668                 } else {
 669                         return $cleansource;
 670                 }
 671         }
 672
 673         /**
 674          * Use the HTML tidy PECL extension to use the tidy library in-process,
 675          * saving the overhead of spawning a new process.
 676          *
 677          * 'pear install tidy' should be able to compile the extension module.
 678          *
 679          * @private
 680          * @static
 681          */
 682         function internalTidy( $text ) {
 683                 global $wgTidyConf, $IP;
 684                 $fname = 'Parser::internalTidy';
 685                 wfProfileIn( $fname );
 686
 687                 $tidy = new tidy;
 688                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 689                 $tidy->cleanRepair();
 690                 if( $tidy->getStatus() == 2 ) {
 691                         // 2 is magic number for fatal error
 692                         // http://www.php.net/manual/en/function.tidy-get-status.php
 693                         $cleansource = null;
 694                 } else {
 695                         $cleansource = tidy_get_output( $tidy );
 696                 }
 697                 wfProfileOut( $fname );
 698                 return $cleansource;
 699         }
 700
 701         /**
 702          * parse the wiki syntax used to render tables
 703          *
 704          * @private
 705          */
 706         function doTableStuff ( $text ) {
 707                 $fname = 'Parser::doTableStuff';
 708                 wfProfileIn( $fname );
 709
 710                 $lines = explode ( "\n" , $text );
 711                 $td_history = array (); // Is currently a td tag open?
 712                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 713                 $tr_history = array (); // Is currently a tr tag open?
 714                 $tr_attributes = array (); // history of tr attributes
 715                 $has_opened_tr = array(); // Did this table open a <tr> element?
 716                 $indent_level = 0; // indent level of the table
 717                 foreach ( $lines as $key => $line )
 718                 {
 719                         $line = trim ( $line );
 720
 721                         if( $line == '' ) { // empty line, go to next line
 722                                 continue;
 723                         }
 724                         $first_character = $line{0};
 725                         $matches = array();
 726
 727                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 728                                 // First check if we are starting a new table
 729                                 $indent_level = strlen( $matches[1] );
 730
 731                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 732                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 733
 734                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 735                                 array_push ( $td_history , false );
 736                                 array_push ( $last_tag_history , '' );
 737                                 array_push ( $tr_history , false );
 738                                 array_push ( $tr_attributes , '' );
 739                                 array_push ( $has_opened_tr , false );
 740                         } else if ( count ( $td_history ) == 0 ) {
 741                                 // Don't do any of the following
 742                                 continue;
 743                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 744                                 // We are ending a table
 745                                 $line = '</table>' . substr ( $line , 2 );
 746                                 $last_tag = array_pop ( $last_tag_history );
 747
 748                                 if ( !array_pop ( $has_opened_tr ) ) {
 749                                         $line = "<tr><td></td></tr>{$line}";
 750                                 }
 751
 752                                 if ( array_pop ( $tr_history ) ) {
 753                                         $line = "</tr>{$line}";
 754                                 }
 755
 756                                 if ( array_pop ( $td_history ) ) {
 757                                         $line = "</{$last_tag}>{$line}";
 758                                 }
 759                                 array_pop ( $tr_attributes );
 760                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 761                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 762                                 // Now we have a table row
 763                                 $line = preg_replace( '#^\|-+#', '', $line );
 764
 765                                 // Whats after the tag is now only attributes
 766                                 $attributes = $this->mStripState->unstripBoth( $line );
 767                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 768                                 array_pop ( $tr_attributes );
 769                                 array_push ( $tr_attributes , $attributes );
 770
 771                                 $line = '';
 772                                 $last_tag = array_pop ( $last_tag_history );
 773                                 array_pop ( $has_opened_tr );
 774                                 array_push ( $has_opened_tr , true );
 775
 776                                 if ( array_pop ( $tr_history ) ) {
 777                                         $line = '</tr>';
 778                                 }
 779
 780                                 if ( array_pop ( $td_history ) ) {
 781                                         $line = "</{$last_tag}>{$line}";
 782                                 }
 783
 784                                 $lines[$key] = $line;
 785                                 array_push ( $tr_history , false );
 786                                 array_push ( $td_history , false );
 787                                 array_push ( $last_tag_history , '' );
 788                         }
 789                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 790                                 // This might be cell elements, td, th or captions
 791                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 792                                         $first_character = '+';
 793                                         $line = substr ( $line , 1 );
 794                                 }
 795
 796                                 $line = substr ( $line , 1 );
 797
 798                                 if ( $first_character == '!' ) {
 799                                         $line = str_replace ( '!!' , '||' , $line );
 800                                 }
 801
 802                                 // Split up multiple cells on the same line.
 803                                 // FIXME : This can result in improper nesting of tags processed
 804                                 // by earlier parser steps, but should avoid splitting up eg
 805                                 // attribute values containing literal "||".
 806                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 807
 808                                 $lines[$key] = '';
 809
 810                                 // Loop through each table cell
 811                                 foreach ( $cells as $cell )
 812                                 {
 813                                         $previous = '';
 814                                         if ( $first_character != '+' )
 815                                         {
 816                                                 $tr_after = array_pop ( $tr_attributes );
 817                                                 if ( !array_pop ( $tr_history ) ) {
 818                                                         $previous = "<tr{$tr_after}>\n";
 819                                                 }
 820                                                 array_push ( $tr_history , true );
 821                                                 array_push ( $tr_attributes , '' );
 822                                                 array_pop ( $has_opened_tr );
 823                                                 array_push ( $has_opened_tr , true );
 824                                         }
 825
 826                                         $last_tag = array_pop ( $last_tag_history );
 827
 828                                         if ( array_pop ( $td_history ) ) {
 829                                                 $previous = "</{$last_tag}>{$previous}";
 830                                         }
 831
 832                                         if ( $first_character == '|' ) {
 833                                                 $last_tag = 'td';
 834                                         } else if ( $first_character == '!' ) {
 835                                                 $last_tag = 'th';
 836                                         } else if ( $first_character == '+' ) {
 837                                                 $last_tag = 'caption';
 838                                         } else {
 839                                                 $last_tag = '';
 840                                         }
 841
 842                                         array_push ( $last_tag_history , $last_tag );
 843
 844                                         // A cell could contain both parameters and data
 845                                         $cell_data = explode ( '|' , $cell , 2 );
 846
 847                                         // Bug 553: Note that a '|' inside an invalid link should not
 848                                         // be mistaken as delimiting cell parameters
 849                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 850                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 851                                         } else if ( count ( $cell_data ) == 1 )
 852                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 853                                         else {
 854                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 855                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 856                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 857                                         }
 858
 859                                         $lines[$key] .= $cell;
 860                                         array_push ( $td_history , true );
 861                                 }
 862                         }
 863                 }
 864
 865                 // Closing open td, tr && table
 866                 while ( count ( $td_history ) > 0 )
 867                 {
 868                         if ( array_pop ( $td_history ) ) {
 869                                 $lines[] = '</td>' ;
 870                         }
 871                         if ( array_pop ( $tr_history ) ) {
 872                                 $lines[] = '</tr>' ;
 873                         }
 874                         if ( !array_pop ( $has_opened_tr ) ) {
 875                                 $lines[] = "<tr><td></td></tr>" ;
 876                         }
 877
 878                         $lines[] = '</table>' ;
 879                 }
 880
 881                 $output = implode ( "\n" , $lines ) ;
 882
 883                 // special case: don't return empty table
 884                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 885                         $output = '';
 886                 }
 887
 888                 wfProfileOut( $fname );
 889
 890                 return $output;
 891         }
 892
 893         /**
 894          * Helper function for parse() that transforms wiki markup into
 895          * HTML. Only called for $mOutputType == OT_HTML.
 896          *
 897          * @private
 898          */
 899         function internalParse( $text ) {
 900                 $isMain = true;
 901                 $fname = 'Parser::internalParse';
 902                 wfProfileIn( $fname );
 903
 904                 # Hook to suspend the parser in this state
 905                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 906                         wfProfileOut( $fname );
 907                         return $text ;
 908                 }
 909
 910                 # Remove <noinclude> tags and <includeonly> sections
 911                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 912                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 913                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 914
 915                 $text = $this->replaceVariables( $text );
 916                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 917                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 918
 919                 // Tables need to come after variable replacement for things to work
 920                 // properly; putting them before other transformations should keep
 921                 // exciting things like link expansions from showing up in surprising
 922                 // places.
 923                 $text = $this->doTableStuff( $text );
 924
 925                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 926
 927                 $text = $this->stripToc( $text );
 928                 $this->stripNoGallery( $text );
 929                 $text = $this->doHeadings( $text );
 930                 if($this->mOptions->getUseDynamicDates()) {
 931                         $df =& DateFormatter::getInstance();
 932                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 933                 }
 934                 $text = $this->doAllQuotes( $text );
 935                 $text = $this->replaceInternalLinks( $text );
 936                 $text = $this->replaceExternalLinks( $text );
 937
 938                 # replaceInternalLinks may sometimes leave behind
 939                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 940                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 941
 942                 $text = $this->doMagicLinks( $text );
 943                 $text = $this->formatHeadings( $text, $isMain );
 944
 945                 wfProfileOut( $fname );
 946                 return $text;
 947         }
 948
 949         /**
 950          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 951          * magic external links.
 952          *
 953          * @private
 954          */
 955         function doMagicLinks( $text ) {
 956                 wfProfileIn( __METHOD__ );
 957                 $text = preg_replace_callback(
 958                         '!(?:                           # Start cases
 959                             <a.*?</a> |                 # Skip link text
 960                             <.*?> |                     # Skip stuff inside HTML elements
 961                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 962                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 963                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 964                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 965                                      [0-9Xx]                 # check digit
 966                                    \b)
 967                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 968                 wfProfileOut( __METHOD__ );
 969                 return $text;
 970         }
 971
 972         function magicLinkCallback( $m ) {
 973                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 974                         # Skip HTML element
 975                         return $m[0];
 976                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 977                         $isbn = $m[2];
 978                         $num = strtr( $isbn, array(
 979                                 '-' => '',
 980                                 ' ' => '',
 981                                 'x' => 'X',
 982                         ));
 983                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
 984                         $text = '<a href="' .
 985                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
 986                                 "\" class=\"internal\">ISBN $isbn</a>";
 987                 } else {
 988                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
 989                                 $keyword = 'RFC';
 990                                 $urlmsg = 'rfcurl';
 991                                 $id = $m[1];
 992                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
 993                                 $keyword = 'PMID';
 994                                 $urlmsg = 'pubmedurl';
 995                                 $id = $m[1];
 996                         } else {
 997                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 998                                         substr($m[0], 0, 20 ) . '"' );
 999                         }
1000
1001                         $url = wfMsg( $urlmsg, $id);
1002                         $sk = $this->mOptions->getSkin();
1003                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1004                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1005                 }
1006                 return $text;
1007         }
1008
1009         /**
1010          * Parse headers and return html
1011          *
1012          * @private
1013          */
1014         function doHeadings( $text ) {
1015                 $fname = 'Parser::doHeadings';
1016                 wfProfileIn( $fname );
1017                 for ( $i = 6; $i >= 1; --$i ) {
1018                         $h = str_repeat( '=', $i );
1019                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1020                           "<h$i>\\1</h$i>", $text );
1021                 }
1022                 wfProfileOut( $fname );
1023                 return $text;
1024         }
1025
1026         /**
1027          * Replace single quotes with HTML markup
1028          * @private
1029          * @return string the altered text
1030          */
1031         function doAllQuotes( $text ) {
1032                 $fname = 'Parser::doAllQuotes';
1033                 wfProfileIn( $fname );
1034                 $outtext = '';
1035                 $lines = explode( "\n", $text );
1036                 foreach ( $lines as $line ) {
1037                         $outtext .= $this->doQuotes ( $line ) . "\n";
1038                 }
1039                 $outtext = substr($outtext, 0,-1);
1040                 wfProfileOut( $fname );
1041                 return $outtext;
1042         }
1043
1044         /**
1045          * Helper function for doAllQuotes()
1046          */
1047         public function doQuotes( $text ) {
1048                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1049                 if ( count( $arr ) == 1 )
1050                         return $text;
1051                 else
1052                 {
1053                         # First, do some preliminary work. This may shift some apostrophes from
1054                         # being mark-up to being text. It also counts the number of occurrences
1055                         # of bold and italics mark-ups.
1056                         $i = 0;
1057                         $numbold = 0;
1058                         $numitalics = 0;
1059                         foreach ( $arr as $r )
1060                         {
1061                                 if ( ( $i % 2 ) == 1 )
1062                                 {
1063                                         # If there are ever four apostrophes, assume the first is supposed to
1064                                         # be text, and the remaining three constitute mark-up for bold text.
1065                                         if ( strlen( $arr[$i] ) == 4 )
1066                                         {
1067                                                 $arr[$i-1] .= "'";
1068                                                 $arr[$i] = "'''";
1069                                         }
1070                                         # If there are more than 5 apostrophes in a row, assume they're all
1071                                         # text except for the last 5.
1072                                         else if ( strlen( $arr[$i] ) > 5 )
1073                                         {
1074                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1075                                                 $arr[$i] = "'''''";
1076                                         }
1077                                         # Count the number of occurrences of bold and italics mark-ups.
1078                                         # We are not counting sequences of five apostrophes.
1079                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1080                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1081                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1082                                 }
1083                                 $i++;
1084                         }
1085
1086                         # If there is an odd number of both bold and italics, it is likely
1087                         # that one of the bold ones was meant to be an apostrophe followed
1088                         # by italics. Which one we cannot know for certain, but it is more
1089                         # likely to be one that has a single-letter word before it.
1090                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1091                         {
1092                                 $i = 0;
1093                                 $firstsingleletterword = -1;
1094                                 $firstmultiletterword = -1;
1095                                 $firstspace = -1;
1096                                 foreach ( $arr as $r )
1097                                 {
1098                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1099                                         {
1100                                                 $x1 = substr ($arr[$i-1], -1);
1101                                                 $x2 = substr ($arr[$i-1], -2, 1);
1102                                                 if ($x1 == ' ') {
1103                                                         if ($firstspace == -1) $firstspace = $i;
1104                                                 } else if ($x2 == ' ') {
1105                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1106                                                 } else {
1107                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1108                                                 }
1109                                         }
1110                                         $i++;
1111                                 }
1112
1113                                 # If there is a single-letter word, use it!
1114                                 if ($firstsingleletterword > -1)
1115                                 {
1116                                         $arr [ $firstsingleletterword ] = "''";
1117                                         $arr [ $firstsingleletterword-1 ] .= "'";
1118                                 }
1119                                 # If not, but there's a multi-letter word, use that one.
1120                                 else if ($firstmultiletterword > -1)
1121                                 {
1122                                         $arr [ $firstmultiletterword ] = "''";
1123                                         $arr [ $firstmultiletterword-1 ] .= "'";
1124                                 }
1125                                 # ... otherwise use the first one that has neither.
1126                                 # (notice that it is possible for all three to be -1 if, for example,
1127                                 # there is only one pentuple-apostrophe in the line)
1128                                 else if ($firstspace > -1)
1129                                 {
1130                                         $arr [ $firstspace ] = "''";
1131                                         $arr [ $firstspace-1 ] .= "'";
1132                                 }
1133                         }
1134
1135                         # Now let's actually convert our apostrophic mush to HTML!
1136                         $output = '';
1137                         $buffer = '';
1138                         $state = '';
1139                         $i = 0;
1140                         foreach ($arr as $r)
1141                         {
1142                                 if (($i % 2) == 0)
1143                                 {
1144                                         if ($state == 'both')
1145                                                 $buffer .= $r;
1146                                         else
1147                                                 $output .= $r;
1148                                 }
1149                                 else
1150                                 {
1151                                         if (strlen ($r) == 2)
1152                                         {
1153                                                 if ($state == 'i')
1154                                                 { $output .= '</i>'; $state = ''; }
1155                                                 else if ($state == 'bi')
1156                                                 { $output .= '</i>'; $state = 'b'; }
1157                                                 else if ($state == 'ib')
1158                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1159                                                 else if ($state == 'both')
1160                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1161                                                 else # $state can be 'b' or ''
1162                                                 { $output .= '<i>'; $state .= 'i'; }
1163                                         }
1164                                         else if (strlen ($r) == 3)
1165                                         {
1166                                                 if ($state == 'b')
1167                                                 { $output .= '</b>'; $state = ''; }
1168                                                 else if ($state == 'bi')
1169                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1170                                                 else if ($state == 'ib')
1171                                                 { $output .= '</b>'; $state = 'i'; }
1172                                                 else if ($state == 'both')
1173                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1174                                                 else # $state can be 'i' or ''
1175                                                 { $output .= '<b>'; $state .= 'b'; }
1176                                         }
1177                                         else if (strlen ($r) == 5)
1178                                         {
1179                                                 if ($state == 'b')
1180                                                 { $output .= '</b><i>'; $state = 'i'; }
1181                                                 else if ($state == 'i')
1182                                                 { $output .= '</i><b>'; $state = 'b'; }
1183                                                 else if ($state == 'bi')
1184                                                 { $output .= '</i></b>'; $state = ''; }
1185                                                 else if ($state == 'ib')
1186                                                 { $output .= '</b></i>'; $state = ''; }
1187                                                 else if ($state == 'both')
1188                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1189                                                 else # ($state == '')
1190                                                 { $buffer = ''; $state = 'both'; }
1191                                         }
1192                                 }
1193                                 $i++;
1194                         }
1195                         # Now close all remaining tags.  Notice that the order is important.
1196                         if ($state == 'b' || $state == 'ib')
1197                                 $output .= '</b>';
1198                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1199                                 $output .= '</i>';
1200                         if ($state == 'bi')
1201                                 $output .= '</b>';
1202                         # There might be lonely ''''', so make sure we have a buffer
1203                         if ($state == 'both' && $buffer)
1204                                 $output .= '<b><i>'.$buffer.'</i></b>';
1205                         return $output;
1206                 }
1207         }
1208
1209         /**
1210          * Replace external links
1211          *
1212          * Note: this is all very hackish and the order of execution matters a lot.
1213          * Make sure to run maintenance/parserTests.php if you change this code.
1214          *
1215          * @private
1216          */
1217         function replaceExternalLinks( $text ) {
1218                 global $wgContLang;
1219                 $fname = 'Parser::replaceExternalLinks';
1220                 wfProfileIn( $fname );
1221
1222                 $sk = $this->mOptions->getSkin();
1223
1224                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1225
1226                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1227
1228                 $i = 0;
1229                 while ( $i<count( $bits ) ) {
1230                         $url = $bits[$i++];
1231                         $protocol = $bits[$i++];
1232                         $text = $bits[$i++];
1233                         $trail = $bits[$i++];
1234
1235                         # The characters '<' and '>' (which were escaped by
1236                         # removeHTMLtags()) should not be included in
1237                         # URLs, per RFC 2396.
1238                         $m2 = array();
1239                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1240                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1241                                 $url = substr($url, 0, $m2[0][1]);
1242                         }
1243
1244                         # If the link text is an image URL, replace it with an <img> tag
1245                         # This happened by accident in the original parser, but some people used it extensively
1246                         $img = $this->maybeMakeExternalImage( $text );
1247                         if ( $img !== false ) {
1248                                 $text = $img;
1249                         }
1250
1251                         $dtrail = '';
1252
1253                         # Set linktype for CSS - if URL==text, link is essentially free
1254                         $linktype = ($text == $url) ? 'free' : 'text';
1255
1256                         # No link text, e.g. [http://domain.tld/some.link]
1257                         if ( $text == '' ) {
1258                                 # Autonumber if allowed. See bug #5918
1259                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1260                                         $text = '[' . ++$this->mAutonumber . ']';
1261                                         $linktype = 'autonumber';
1262                                 } else {
1263                                         # Otherwise just use the URL
1264                                         $text = htmlspecialchars( $url );
1265                                         $linktype = 'free';
1266                                 }
1267                         } else {
1268                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1269                                 # Check for trail
1270                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1271                         }
1272
1273                         $text = $wgContLang->markNoConversion($text);
1274
1275                         $url = Sanitizer::cleanUrl( $url );
1276
1277                         # Process the trail (i.e. everything after this link up until start of the next link),
1278                         # replacing any non-bracketed links
1279                         $trail = $this->replaceFreeExternalLinks( $trail );
1280
1281                         # Use the encoded URL
1282                         # This means that users can paste URLs directly into the text
1283                         # Funny characters like &ouml; aren't valid in URLs anyway
1284                         # This was changed in August 2004
1285                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1286
1287                         # Register link in the output object.
1288                         # Replace unnecessary URL escape codes with the referenced character
1289                         # This prevents spammers from hiding links from the filters
1290                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1291                         $this->mOutput->addExternalLink( $pasteurized );
1292                 }
1293
1294                 wfProfileOut( $fname );
1295                 return $s;
1296         }
1297
1298         /**
1299          * Replace anything that looks like a URL with a link
1300          * @private
1301          */
1302         function replaceFreeExternalLinks( $text ) {
1303                 global $wgContLang;
1304                 $fname = 'Parser::replaceFreeExternalLinks';
1305                 wfProfileIn( $fname );
1306
1307                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1308                 $s = array_shift( $bits );
1309                 $i = 0;
1310
1311                 $sk = $this->mOptions->getSkin();
1312
1313                 while ( $i < count( $bits ) ){
1314                         $protocol = $bits[$i++];
1315                         $remainder = $bits[$i++];
1316
1317                         $m = array();
1318                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1319                                 # Found some characters after the protocol that look promising
1320                                 $url = $protocol . $m[1];
1321                                 $trail = $m[2];
1322
1323                                 # special case: handle urls as url args:
1324                                 # http://www.example.com/foo?=http://www.example.com/bar
1325                                 if(strlen($trail) == 0 &&
1326                                         isset($bits[$i]) &&
1327                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1328                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1329                                 {
1330                                         # add protocol, arg
1331                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1332                                         $i += 2;
1333                                         $trail = $m[2];
1334                                 }
1335
1336                                 # The characters '<' and '>' (which were escaped by
1337                                 # removeHTMLtags()) should not be included in
1338                                 # URLs, per RFC 2396.
1339                                 $m2 = array();
1340                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1341                                         $trail = substr($url, $m2[0][1]) . $trail;
1342                                         $url = substr($url, 0, $m2[0][1]);
1343                                 }
1344
1345                                 # Move trailing punctuation to $trail
1346                                 $sep = ',;\.:!?';
1347                                 # If there is no left bracket, then consider right brackets fair game too
1348                                 if ( strpos( $url, '(' ) === false ) {
1349                                         $sep .= ')';
1350                                 }
1351
1352                                 $numSepChars = strspn( strrev( $url ), $sep );
1353                                 if ( $numSepChars ) {
1354                                         $trail = substr( $url, -$numSepChars ) . $trail;
1355                                         $url = substr( $url, 0, -$numSepChars );
1356                                 }
1357
1358                                 $url = Sanitizer::cleanUrl( $url );
1359
1360                                 # Is this an external image?
1361                                 $text = $this->maybeMakeExternalImage( $url );
1362                                 if ( $text === false ) {
1363                                         # Not an image, make a link
1364                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1365                                         # Register it in the output object...
1366                                         # Replace unnecessary URL escape codes with their equivalent characters
1367                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1368                                         $this->mOutput->addExternalLink( $pasteurized );
1369                                 }
1370                                 $s .= $text . $trail;
1371                         } else {
1372                                 $s .= $protocol . $remainder;
1373                         }
1374                 }
1375                 wfProfileOut( $fname );
1376                 return $s;
1377         }
1378
1379         /**
1380          * Replace unusual URL escape codes with their equivalent characters
1381          * @param string
1382          * @return string
1383          * @static
1384          * @todo  This can merge genuinely required bits in the path or query string,
1385          *        breaking legit URLs. A proper fix would treat the various parts of
1386          *        the URL differently; as a workaround, just use the output for
1387          *        statistical records, not for actual linking/output.
1388          */
1389         static function replaceUnusualEscapes( $url ) {
1390                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1391                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1392         }
1393
1394         /**
1395          * Callback function used in replaceUnusualEscapes().
1396          * Replaces unusual URL escape codes with their equivalent character
1397          * @static
1398          * @private
1399          */
1400         private static function replaceUnusualEscapesCallback( $matches ) {
1401                 $char = urldecode( $matches[0] );
1402                 $ord = ord( $char );
1403                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1404                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1405                         // No, shouldn't be escaped
1406                         return $char;
1407                 } else {
1408                         // Yes, leave it escaped
1409                         return $matches[0];
1410                 }
1411         }
1412
1413         /**
1414          * make an image if it's allowed, either through the global
1415          * option or through the exception
1416          * @private
1417          */
1418         function maybeMakeExternalImage( $url ) {
1419                 $sk = $this->mOptions->getSkin();
1420                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1421                 $imagesexception = !empty($imagesfrom);
1422                 $text = false;
1423                 if ( $this->mOptions->getAllowExternalImages()
1424                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1425                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1426                                 # Image found
1427                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1428                         }
1429                 }
1430                 return $text;
1431         }
1432
1433         /**
1434          * Process [[ ]] wikilinks
1435          *
1436          * @private
1437          */
1438         function replaceInternalLinks( $s ) {
1439                 global $wgContLang;
1440                 static $fname = 'Parser::replaceInternalLinks' ;
1441
1442                 wfProfileIn( $fname );
1443
1444                 wfProfileIn( $fname.'-setup' );
1445                 static $tc = FALSE;
1446                 # the % is needed to support urlencoded titles as well
1447                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1448
1449                 $sk = $this->mOptions->getSkin();
1450
1451                 #split the entire text string on occurences of [[
1452                 $a = explode( '[[', ' ' . $s );
1453                 #get the first element (all text up to first [[), and remove the space we added
1454                 $s = array_shift( $a );
1455                 $s = substr( $s, 1 );
1456
1457                 # Match a link having the form [[namespace:link|alternate]]trail
1458                 static $e1 = FALSE;
1459                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1460                 # Match cases where there is no "]]", which might still be images
1461                 static $e1_img = FALSE;
1462                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1463
1464                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1465                 $e2 = null;
1466                 if ( $useLinkPrefixExtension ) {
1467                         # Match the end of a line for a word that's not followed by whitespace,
1468                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1469                         $e2 = wfMsgForContent( 'linkprefix' );
1470                 }
1471
1472                 if( is_null( $this->mTitle ) ) {
1473                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1474                 }
1475                 $nottalk = !$this->mTitle->isTalkPage();
1476
1477                 if ( $useLinkPrefixExtension ) {
1478                         $m = array();
1479                         if ( preg_match( $e2, $s, $m ) ) {
1480                                 $first_prefix = $m[2];
1481                         } else {
1482                                 $first_prefix = false;
1483                         }
1484                 } else {
1485                         $prefix = '';
1486                 }
1487
1488                 if($wgContLang->hasVariants()) {
1489                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1490                 } else {
1491                         $selflink = array($this->mTitle->getPrefixedText());
1492                 }
1493                 $useSubpages = $this->areSubpagesAllowed();
1494                 wfProfileOut( $fname.'-setup' );
1495
1496                 # Loop for each link
1497                 for ($k = 0; isset( $a[$k] ); $k++) {
1498                         $line = $a[$k];
1499                         if ( $useLinkPrefixExtension ) {
1500                                 wfProfileIn( $fname.'-prefixhandling' );
1501                                 if ( preg_match( $e2, $s, $m ) ) {
1502                                         $prefix = $m[2];
1503                                         $s = $m[1];
1504                                 } else {
1505                                         $prefix='';
1506                                 }
1507                                 # first link
1508                                 if($first_prefix) {
1509                                         $prefix = $first_prefix;
1510                                         $first_prefix = false;
1511                                 }
1512                                 wfProfileOut( $fname.'-prefixhandling' );
1513                         }
1514
1515                         $might_be_img = false;
1516
1517                         wfProfileIn( "$fname-e1" );
1518                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1519                                 $text = $m[2];
1520                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1521                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1522                                 # the real problem is with the $e1 regex
1523                                 # See bug 1300.
1524                                 #
1525                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1526                                 # and no image is in sight. See bug 2095.
1527                                 #
1528                                 if( $text !== '' &&
1529                                         substr( $m[3], 0, 1 ) === ']' &&
1530                                         strpos($text, '[') !== false
1531                                 )
1532                                 {
1533                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1534                                         $m[3] = substr( $m[3], 1 );
1535                                 }
1536                                 # fix up urlencoded title texts
1537                                 if( strpos( $m[1], '%' ) !== false ) {
1538                                         # Should anchors '#' also be rejected?
1539                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1540                                 }
1541                                 $trail = $m[3];
1542                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1543                                 $might_be_img = true;
1544                                 $text = $m[2];
1545                                 if ( strpos( $m[1], '%' ) !== false ) {
1546                                         $m[1] = urldecode($m[1]);
1547                                 }
1548                                 $trail = "";
1549                         } else { # Invalid form; output directly
1550                                 $s .= $prefix . '[[' . $line ;
1551                                 wfProfileOut( "$fname-e1" );
1552                                 continue;
1553                         }
1554                         wfProfileOut( "$fname-e1" );
1555                         wfProfileIn( "$fname-misc" );
1556
1557                         # Don't allow internal links to pages containing
1558                         # PROTO: where PROTO is a valid URL protocol; these
1559                         # should be external links.
1560                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1561                                 $s .= $prefix . '[[' . $line ;
1562                                 continue;
1563                         }
1564
1565                         # Make subpage if necessary
1566                         if( $useSubpages ) {
1567                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1568                         } else {
1569                                 $link = $m[1];
1570                         }
1571
1572                         $noforce = (substr($m[1], 0, 1) != ':');
1573                         if (!$noforce) {
1574                                 # Strip off leading ':'
1575                                 $link = substr($link, 1);
1576                         }
1577
1578                         wfProfileOut( "$fname-misc" );
1579                         wfProfileIn( "$fname-title" );
1580                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1581                         if( !$nt ) {
1582                                 $s .= $prefix . '[[' . $line;
1583                                 wfProfileOut( "$fname-title" );
1584                                 continue;
1585                         }
1586
1587                         $ns = $nt->getNamespace();
1588                         $iw = $nt->getInterWiki();
1589                         wfProfileOut( "$fname-title" );
1590
1591                         if ($might_be_img) { # if this is actually an invalid link
1592                                 wfProfileIn( "$fname-might_be_img" );
1593                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1594                                         $found = false;
1595                                         while (isset ($a[$k+1]) ) {
1596                                                 #look at the next 'line' to see if we can close it there
1597                                                 $spliced = array_splice( $a, $k + 1, 1 );
1598                                                 $next_line = array_shift( $spliced );
1599                                                 $m = explode( ']]', $next_line, 3 );
1600                                                 if ( count( $m ) == 3 ) {
1601                                                         # the first ]] closes the inner link, the second the image
1602                                                         $found = true;
1603                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1604                                                         $trail = $m[2];
1605                                                         break;
1606                                                 } elseif ( count( $m ) == 2 ) {
1607                                                         #if there's exactly one ]] that's fine, we'll keep looking
1608                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1609                                                 } else {
1610                                                         #if $next_line is invalid too, we need look no further
1611                                                         $text .= '[[' . $next_line;
1612                                                         break;
1613                                                 }
1614                                         }
1615                                         if ( !$found ) {
1616                                                 # we couldn't find the end of this imageLink, so output it raw
1617                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1618                                                 $text = $this->replaceInternalLinks($text);
1619                                                 $s .= "{$prefix}[[$link|$text";
1620                                                 # note: no $trail, because without an end, there *is* no trail
1621                                                 wfProfileOut( "$fname-might_be_img" );
1622                                                 continue;
1623                                         }
1624                                 } else { #it's not an image, so output it raw
1625                                         $s .= "{$prefix}[[$link|$text";
1626                                         # note: no $trail, because without an end, there *is* no trail
1627                                         wfProfileOut( "$fname-might_be_img" );
1628                                         continue;
1629                                 }
1630                                 wfProfileOut( "$fname-might_be_img" );
1631                         }
1632
1633                         $wasblank = ( '' == $text );
1634                         if( $wasblank ) $text = $link;
1635
1636                         # Link not escaped by : , create the various objects
1637                         if( $noforce ) {
1638
1639                                 # Interwikis
1640                                 wfProfileIn( "$fname-interwiki" );
1641                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1642                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1643                                         $s = rtrim($s . $prefix);
1644                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1645                                         wfProfileOut( "$fname-interwiki" );
1646                                         continue;
1647                                 }
1648                                 wfProfileOut( "$fname-interwiki" );
1649
1650                                 if ( $ns == NS_IMAGE ) {
1651                                         wfProfileIn( "$fname-image" );
1652                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1653                                                 # recursively parse links inside the image caption
1654                                                 # actually, this will parse them in any other parameters, too,
1655                                                 # but it might be hard to fix that, and it doesn't matter ATM
1656                                                 $text = $this->replaceExternalLinks($text);
1657                                                 $text = $this->replaceInternalLinks($text);
1658
1659                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1660                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1661                                                 $this->mOutput->addImage( $nt->getDBkey() );
1662
1663                                                 wfProfileOut( "$fname-image" );
1664                                                 continue;
1665                                         } else {
1666                                                 # We still need to record the image's presence on the page
1667                                                 $this->mOutput->addImage( $nt->getDBkey() );
1668                                         }
1669                                         wfProfileOut( "$fname-image" );
1670
1671                                 }
1672
1673                                 if ( $ns == NS_CATEGORY ) {
1674                                         wfProfileIn( "$fname-category" );
1675                                         $s = rtrim($s . "\n"); # bug 87
1676
1677                                         if ( $wasblank ) {
1678                                                 $sortkey = $this->getDefaultSort();
1679                                         } else {
1680                                                 $sortkey = $text;
1681                                         }
1682                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1683                                         $sortkey = str_replace( "\n", '', $sortkey );
1684                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1685                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1686
1687                                         /**
1688                                          * Strip the whitespace Category links produce, see bug 87
1689                                          * @todo We might want to use trim($tmp, "\n") here.
1690                                          */
1691                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1692
1693                                         wfProfileOut( "$fname-category" );
1694                                         continue;
1695                                 }
1696                         }
1697
1698                         # Self-link checking
1699                         if( $nt->getFragment() === '' ) {
1700                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1701                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1702                                         continue;
1703                                 }
1704                         }
1705
1706                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1707                         if( $ns == NS_MEDIA ) {
1708                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1709                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1710                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1711                                 $this->mOutput->addImage( $nt->getDBkey() );
1712                                 continue;
1713                         } elseif( $ns == NS_SPECIAL ) {
1714                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1715                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1716                                 } else {
1717                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1718                                 }
1719                                 continue;
1720                         } elseif( $ns == NS_IMAGE ) {
1721                                 $img = wfFindFile( $nt );
1722                                 if( $img ) {
1723                                         // Force a blue link if the file exists; may be a remote
1724                                         // upload on the shared repository, and we want to see its
1725                                         // auto-generated page.
1726                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1727                                         $this->mOutput->addLink( $nt );
1728                                         continue;
1729                                 }
1730                         }
1731                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1732                 }
1733                 wfProfileOut( $fname );
1734                 return $s;
1735         }
1736
1737         /**
1738          * Make a link placeholder. The text returned can be later resolved to a real link with
1739          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1740          * parsing of interwiki links, and secondly to allow all existence checks and
1741          * article length checks (for stub links) to be bundled into a single query.
1742          *
1743          */
1744         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1745                 wfProfileIn( __METHOD__ );
1746                 if ( ! is_object($nt) ) {
1747                         # Fail gracefully
1748                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1749                 } else {
1750                         # Separate the link trail from the rest of the link
1751                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1752
1753                         if ( $nt->isExternal() ) {
1754                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1755                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1756                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1757                         } else {
1758                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1759                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1760                                 $this->mLinkHolders['queries'][] = $query;
1761                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1762                                 $this->mLinkHolders['titles'][] = $nt;
1763
1764                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1765                         }
1766                 }
1767                 wfProfileOut( __METHOD__ );
1768                 return $retVal;
1769         }
1770
1771         /**
1772          * Render a forced-blue link inline; protect against double expansion of
1773          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1774          * Since this little disaster has to split off the trail text to avoid
1775          * breaking URLs in the following text without breaking trails on the
1776          * wiki links, it's been made into a horrible function.
1777          *
1778          * @param Title $nt
1779          * @param string $text
1780          * @param string $query
1781          * @param string $trail
1782          * @param string $prefix
1783          * @return string HTML-wikitext mix oh yuck
1784          */
1785         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1786                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1787                 $sk = $this->mOptions->getSkin();
1788                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1789                 return $this->armorLinks( $link ) . $trail;
1790         }
1791
1792         /**
1793          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1794          * going to go through further parsing steps before inline URL expansion.
1795          *
1796          * In particular this is important when using action=render, which causes
1797          * full URLs to be included.
1798          *
1799          * Oh man I hate our multi-layer parser!
1800          *
1801          * @param string more-or-less HTML
1802          * @return string less-or-more HTML with NOPARSE bits
1803          */
1804         function armorLinks( $text ) {
1805                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1806                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1807         }
1808
1809         /**
1810          * Return true if subpage links should be expanded on this page.
1811          * @return bool
1812          */
1813         function areSubpagesAllowed() {
1814                 # Some namespaces don't allow subpages
1815                 global $wgNamespacesWithSubpages;
1816                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1817         }
1818
1819         /**
1820          * Handle link to subpage if necessary
1821          * @param string $target the source of the link
1822          * @param string &$text the link text, modified as necessary
1823          * @return string the full name of the link
1824          * @private
1825          */
1826         function maybeDoSubpageLink($target, &$text) {
1827                 # Valid link forms:
1828                 # Foobar -- normal
1829                 # :Foobar -- override special treatment of prefix (images, language links)
1830                 # /Foobar -- convert to CurrentPage/Foobar
1831                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1832                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1833                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1834
1835                 $fname = 'Parser::maybeDoSubpageLink';
1836                 wfProfileIn( $fname );
1837                 $ret = $target; # default return value is no change
1838
1839                 # Some namespaces don't allow subpages,
1840                 # so only perform processing if subpages are allowed
1841                 if( $this->areSubpagesAllowed() ) {
1842                         $hash = strpos( $target, '#' );
1843                         if( $hash !== false ) {
1844                                 $suffix = substr( $target, $hash );
1845                                 $target = substr( $target, 0, $hash );
1846                         } else {
1847                                 $suffix = '';
1848                         }
1849                         # bug 7425
1850                         $target = trim( $target );
1851                         # Look at the first character
1852                         if( $target != '' && $target{0} == '/' ) {
1853                                 # / at end means we don't want the slash to be shown
1854                                 $m = array();
1855                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1856                                 if( $trailingSlashes ) {
1857                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1858                                 } else {
1859                                         $noslash = substr( $target, 1 );
1860                                 }
1861
1862                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1863                                 if( '' === $text ) {
1864                                         $text = $target . $suffix;
1865                                 } # this might be changed for ugliness reasons
1866                         } else {
1867                                 # check for .. subpage backlinks
1868                                 $dotdotcount = 0;
1869                                 $nodotdot = $target;
1870                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1871                                         ++$dotdotcount;
1872                                         $nodotdot = substr( $nodotdot, 3 );
1873                                 }
1874                                 if($dotdotcount > 0) {
1875                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1876                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1877                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1878                                                 # / at the end means don't show full path
1879                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1880                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1881                                                         if( '' === $text ) {
1882                                                                 $text = $nodotdot . $suffix;
1883                                                         }
1884                                                 }
1885                                                 $nodotdot = trim( $nodotdot );
1886                                                 if( $nodotdot != '' ) {
1887                                                         $ret .= '/' . $nodotdot;
1888                                                 }
1889                                                 $ret .= $suffix;
1890                                         }
1891                                 }
1892                         }
1893                 }
1894
1895                 wfProfileOut( $fname );
1896                 return $ret;
1897         }
1898
1899         /**#@+
1900          * Used by doBlockLevels()
1901          * @private
1902          */
1903         /* private */ function closeParagraph() {
1904                 $result = '';
1905                 if ( '' != $this->mLastSection ) {
1906                         $result = '</' . $this->mLastSection  . ">\n";
1907                 }
1908                 $this->mInPre = false;
1909                 $this->mLastSection = '';
1910                 return $result;
1911         }
1912         # getCommon() returns the length of the longest common substring
1913         # of both arguments, starting at the beginning of both.
1914         #
1915         /* private */ function getCommon( $st1, $st2 ) {
1916                 $fl = strlen( $st1 );
1917                 $shorter = strlen( $st2 );
1918                 if ( $fl < $shorter ) { $shorter = $fl; }
1919
1920                 for ( $i = 0; $i < $shorter; ++$i ) {
1921                         if ( $st1{$i} != $st2{$i} ) { break; }
1922                 }
1923                 return $i;
1924         }
1925         # These next three functions open, continue, and close the list
1926         # element appropriate to the prefix character passed into them.
1927         #
1928         /* private */ function openList( $char ) {
1929                 $result = $this->closeParagraph();
1930
1931                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1932                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1933                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1934                 else if ( ';' == $char ) {
1935                         $result .= '<dl><dt>';
1936                         $this->mDTopen = true;
1937                 }
1938                 else { $result = '<!-- ERR 1 -->'; }
1939
1940                 return $result;
1941         }
1942
1943         /* private */ function nextItem( $char ) {
1944                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1945                 else if ( ':' == $char || ';' == $char ) {
1946                         $close = '</dd>';
1947                         if ( $this->mDTopen ) { $close = '</dt>'; }
1948                         if ( ';' == $char ) {
1949                                 $this->mDTopen = true;
1950                                 return $close . '<dt>';
1951                         } else {
1952                                 $this->mDTopen = false;
1953                                 return $close . '<dd>';
1954                         }
1955                 }
1956                 return '<!-- ERR 2 -->';
1957         }
1958
1959         /* private */ function closeList( $char ) {
1960                 if ( '*' == $char ) { $text = '</li></ul>'; }
1961                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1962                 else if ( ':' == $char ) {
1963                         if ( $this->mDTopen ) {
1964                                 $this->mDTopen = false;
1965                                 $text = '</dt></dl>';
1966                         } else {
1967                                 $text = '</dd></dl>';
1968                         }
1969                 }
1970                 else {  return '<!-- ERR 3 -->'; }
1971                 return $text."\n";
1972         }
1973         /**#@-*/
1974
1975         /**
1976          * Make lists from lines starting with ':', '*', '#', etc.
1977          *
1978          * @private
1979          * @return string the lists rendered as HTML
1980          */
1981         function doBlockLevels( $text, $linestart ) {
1982                 $fname = 'Parser::doBlockLevels';
1983                 wfProfileIn( $fname );
1984
1985                 # Parsing through the text line by line.  The main thing
1986                 # happening here is handling of block-level elements p, pre,
1987                 # and making lists from lines starting with * # : etc.
1988                 #
1989                 $textLines = explode( "\n", $text );
1990
1991                 $lastPrefix = $output = '';
1992                 $this->mDTopen = $inBlockElem = false;
1993                 $prefixLength = 0;
1994                 $paragraphStack = false;
1995
1996                 if ( !$linestart ) {
1997                         $output .= array_shift( $textLines );
1998                 }
1999                 foreach ( $textLines as $oLine ) {
2000                         $lastPrefixLength = strlen( $lastPrefix );
2001                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2002                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2003                         if ( !$this->mInPre ) {
2004                                 # Multiple prefixes may abut each other for nested lists.
2005                                 $prefixLength = strspn( $oLine, '*#:;' );
2006                                 $pref = substr( $oLine, 0, $prefixLength );
2007
2008                                 # eh?
2009                                 $pref2 = str_replace( ';', ':', $pref );
2010                                 $t = substr( $oLine, $prefixLength );
2011                                 $this->mInPre = !empty($preOpenMatch);
2012                         } else {
2013                                 # Don't interpret any other prefixes in preformatted text
2014                                 $prefixLength = 0;
2015                                 $pref = $pref2 = '';
2016                                 $t = $oLine;
2017                         }
2018
2019                         # List generation
2020                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2021                                 # Same as the last item, so no need to deal with nesting or opening stuff
2022                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2023                                 $paragraphStack = false;
2024
2025                                 if ( substr( $pref, -1 ) == ';') {
2026                                         # The one nasty exception: definition lists work like this:
2027                                         # ; title : definition text
2028                                         # So we check for : in the remainder text to split up the
2029                                         # title and definition, without b0rking links.
2030                                         $term = $t2 = '';
2031                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2032                                                 $t = $t2;
2033                                                 $output .= $term . $this->nextItem( ':' );
2034                                         }
2035                                 }
2036                         } elseif( $prefixLength || $lastPrefixLength ) {
2037                                 # Either open or close a level...
2038                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2039                                 $paragraphStack = false;
2040
2041                                 while( $commonPrefixLength < $lastPrefixLength ) {
2042                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2043                                         --$lastPrefixLength;
2044                                 }
2045                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2046                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2047                                 }
2048                                 while ( $prefixLength > $commonPrefixLength ) {
2049                                         $char = substr( $pref, $commonPrefixLength, 1 );
2050                                         $output .= $this->openList( $char );
2051
2052                                         if ( ';' == $char ) {
2053                                                 # FIXME: This is dupe of code above
2054                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2055                                                         $t = $t2;
2056                                                         $output .= $term . $this->nextItem( ':' );
2057                                                 }
2058                                         }
2059                                         ++$commonPrefixLength;
2060                                 }
2061                                 $lastPrefix = $pref2;
2062                         }
2063                         if( 0 == $prefixLength ) {
2064                                 wfProfileIn( "$fname-paragraph" );
2065                                 # No prefix (not in list)--go to paragraph mode
2066                                 // XXX: use a stack for nestable elements like span, table and div
2067                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2068                                 $closematch = preg_match(
2069                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2070                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2071                                 if ( $openmatch or $closematch ) {
2072                                         $paragraphStack = false;
2073                                         # TODO bug 5718: paragraph closed
2074                                         $output .= $this->closeParagraph();
2075                                         if ( $preOpenMatch and !$preCloseMatch ) {
2076                                                 $this->mInPre = true;
2077                                         }
2078                                         if ( $closematch ) {
2079                                                 $inBlockElem = false;
2080                                         } else {
2081                                                 $inBlockElem = true;
2082                                         }
2083                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2084                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2085                                                 // pre
2086                                                 if ($this->mLastSection != 'pre') {
2087                                                         $paragraphStack = false;
2088                                                         $output .= $this->closeParagraph().'<pre>';
2089                                                         $this->mLastSection = 'pre';
2090                                                 }
2091                                                 $t = substr( $t, 1 );
2092                                         } else {
2093                                                 // paragraph
2094                                                 if ( '' == trim($t) ) {
2095                                                         if ( $paragraphStack ) {
2096                                                                 $output .= $paragraphStack.'<br />';
2097                                                                 $paragraphStack = false;
2098                                                                 $this->mLastSection = 'p';
2099                                                         } else {
2100                                                                 if ($this->mLastSection != 'p' ) {
2101                                                                         $output .= $this->closeParagraph();
2102                                                                         $this->mLastSection = '';
2103                                                                         $paragraphStack = '<p>';
2104                                                                 } else {
2105                                                                         $paragraphStack = '</p><p>';
2106                                                                 }
2107                                                         }
2108                                                 } else {
2109                                                         if ( $paragraphStack ) {
2110                                                                 $output .= $paragraphStack;
2111                                                                 $paragraphStack = false;
2112                                                                 $this->mLastSection = 'p';
2113                                                         } else if ($this->mLastSection != 'p') {
2114                                                                 $output .= $this->closeParagraph().'<p>';
2115                                                                 $this->mLastSection = 'p';
2116                                                         }
2117                                                 }
2118                                         }
2119                                 }
2120                                 wfProfileOut( "$fname-paragraph" );
2121                         }
2122                         // somewhere above we forget to get out of pre block (bug 785)
2123                         if($preCloseMatch && $this->mInPre) {
2124                                 $this->mInPre = false;
2125                         }
2126                         if ($paragraphStack === false) {
2127                                 $output .= $t."\n";
2128                         }
2129                 }
2130                 while ( $prefixLength ) {
2131                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2132                         --$prefixLength;
2133                 }
2134                 if ( '' != $this->mLastSection ) {
2135                         $output .= '</' . $this->mLastSection . '>';
2136                         $this->mLastSection = '';
2137                 }
2138
2139                 wfProfileOut( $fname );
2140                 return $output;
2141         }
2142
2143         /**
2144          * Split up a string on ':', ignoring any occurences inside tags
2145          * to prevent illegal overlapping.
2146          * @param string $str the string to split
2147          * @param string &$before set to everything before the ':'
2148          * @param string &$after set to everything after the ':'
2149          * return string the position of the ':', or false if none found
2150          */
2151         function findColonNoLinks($str, &$before, &$after) {
2152                 $fname = 'Parser::findColonNoLinks';
2153                 wfProfileIn( $fname );
2154
2155                 $pos = strpos( $str, ':' );
2156                 if( $pos === false ) {
2157                         // Nothing to find!
2158                         wfProfileOut( $fname );
2159                         return false;
2160                 }
2161
2162                 $lt = strpos( $str, '<' );
2163                 if( $lt === false || $lt > $pos ) {
2164                         // Easy; no tag nesting to worry about
2165                         $before = substr( $str, 0, $pos );
2166                         $after = substr( $str, $pos+1 );
2167                         wfProfileOut( $fname );
2168                         return $pos;
2169                 }
2170
2171                 // Ugly state machine to walk through avoiding tags.
2172                 $state = self::COLON_STATE_TEXT;
2173                 $stack = 0;
2174                 $len = strlen( $str );
2175                 for( $i = 0; $i < $len; $i++ ) {
2176                         $c = $str{$i};
2177
2178                         switch( $state ) {
2179                         // (Using the number is a performance hack for common cases)
2180                         case 0: // self::COLON_STATE_TEXT:
2181                                 switch( $c ) {
2182                                 case "<":
2183                                         // Could be either a <start> tag or an </end> tag
2184                                         $state = self::COLON_STATE_TAGSTART;
2185                                         break;
2186                                 case ":":
2187                                         if( $stack == 0 ) {
2188                                                 // We found it!
2189                                                 $before = substr( $str, 0, $i );
2190                                                 $after = substr( $str, $i + 1 );
2191                                                 wfProfileOut( $fname );
2192                                                 return $i;
2193                                         }
2194                                         // Embedded in a tag; don't break it.
2195                                         break;
2196                                 default:
2197                                         // Skip ahead looking for something interesting
2198                                         $colon = strpos( $str, ':', $i );
2199                                         if( $colon === false ) {
2200                                                 // Nothing else interesting
2201                                                 wfProfileOut( $fname );
2202                                                 return false;
2203                                         }
2204                                         $lt = strpos( $str, '<', $i );
2205                                         if( $stack === 0 ) {
2206                                                 if( $lt === false || $colon < $lt ) {
2207                                                         // We found it!
2208                                                         $before = substr( $str, 0, $colon );
2209                                                         $after = substr( $str, $colon + 1 );
2210                                                         wfProfileOut( $fname );
2211                                                         return $i;
2212                                                 }
2213                                         }
2214                                         if( $lt === false ) {
2215                                                 // Nothing else interesting to find; abort!
2216                                                 // We're nested, but there's no close tags left. Abort!
2217                                                 break 2;
2218                                         }
2219                                         // Skip ahead to next tag start
2220                                         $i = $lt;
2221                                         $state = self::COLON_STATE_TAGSTART;
2222                                 }
2223                                 break;
2224                         case 1: // self::COLON_STATE_TAG:
2225                                 // In a <tag>
2226                                 switch( $c ) {
2227                                 case ">":
2228                                         $stack++;
2229                                         $state = self::COLON_STATE_TEXT;
2230                                         break;
2231                                 case "/":
2232                                         // Slash may be followed by >?
2233                                         $state = self::COLON_STATE_TAGSLASH;
2234                                         break;
2235                                 default:
2236                                         // ignore
2237                                 }
2238                                 break;
2239                         case 2: // self::COLON_STATE_TAGSTART:
2240                                 switch( $c ) {
2241                                 case "/":
2242                                         $state = self::COLON_STATE_CLOSETAG;
2243                                         break;
2244                                 case "!":
2245                                         $state = self::COLON_STATE_COMMENT;
2246                                         break;
2247                                 case ">":
2248                                         // Illegal early close? This shouldn't happen D:
2249                                         $state = self::COLON_STATE_TEXT;
2250                                         break;
2251                                 default:
2252                                         $state = self::COLON_STATE_TAG;
2253                                 }
2254                                 break;
2255                         case 3: // self::COLON_STATE_CLOSETAG:
2256                                 // In a </tag>
2257                                 if( $c == ">" ) {
2258                                         $stack--;
2259                                         if( $stack < 0 ) {
2260                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2261                                                 wfProfileOut( $fname );
2262                                                 return false;
2263                                         }
2264                                         $state = self::COLON_STATE_TEXT;
2265                                 }
2266                                 break;
2267                         case self::COLON_STATE_TAGSLASH:
2268                                 if( $c == ">" ) {
2269                                         // Yes, a self-closed tag <blah/>
2270                                         $state = self::COLON_STATE_TEXT;
2271                                 } else {
2272                                         // Probably we're jumping the gun, and this is an attribute
2273                                         $state = self::COLON_STATE_TAG;
2274                                 }
2275                                 break;
2276                         case 5: // self::COLON_STATE_COMMENT:
2277                                 if( $c == "-" ) {
2278                                         $state = self::COLON_STATE_COMMENTDASH;
2279                                 }
2280                                 break;
2281                         case self::COLON_STATE_COMMENTDASH:
2282                                 if( $c == "-" ) {
2283                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2284                                 } else {
2285                                         $state = self::COLON_STATE_COMMENT;
2286                                 }
2287                                 break;
2288                         case self::COLON_STATE_COMMENTDASHDASH:
2289                                 if( $c == ">" ) {
2290                                         $state = self::COLON_STATE_TEXT;
2291                                 } else {
2292                                         $state = self::COLON_STATE_COMMENT;
2293                                 }
2294                                 break;
2295                         default:
2296                                 throw new MWException( "State machine error in $fname" );
2297                         }
2298                 }
2299                 if( $stack > 0 ) {
2300                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2301                         return false;
2302                 }
2303                 wfProfileOut( $fname );
2304                 return false;
2305         }
2306
2307         /**
2308          * Return value of a magic variable (like PAGENAME)
2309          *
2310          * @private
2311          */
2312         function getVariableValue( $index ) {
2313                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2314
2315                 /**
2316                  * Some of these require message or data lookups and can be
2317                  * expensive to check many times.
2318                  */
2319                 static $varCache = array();
2320                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2321                         if ( isset( $varCache[$index] ) ) {
2322                                 return $varCache[$index];
2323                         }
2324                 }
2325
2326                 $ts = time();
2327                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2328
2329                 # Use the time zone
2330                 global $wgLocaltimezone;
2331                 if ( isset( $wgLocaltimezone ) ) {
2332                         $oldtz = getenv( 'TZ' );
2333                         putenv( 'TZ='.$wgLocaltimezone );
2334                 }
2335
2336                 wfSuppressWarnings(); // E_STRICT system time bitching
2337                 $localTimestamp = date( 'YmdHis', $ts );
2338                 $localMonth = date( 'm', $ts );
2339                 $localMonthName = date( 'n', $ts );
2340                 $localDay = date( 'j', $ts );
2341                 $localDay2 = date( 'd', $ts );
2342                 $localDayOfWeek = date( 'w', $ts );
2343                 $localWeek = date( 'W', $ts );
2344                 $localYear = date( 'Y', $ts );
2345                 $localHour = date( 'H', $ts );
2346                 if ( isset( $wgLocaltimezone ) ) {
2347                         putenv( 'TZ='.$oldtz );
2348                 }
2349                 wfRestoreWarnings();
2350
2351                 switch ( $index ) {
2352                         case 'currentmonth':
2353                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2354                         case 'currentmonthname':
2355                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2356                         case 'currentmonthnamegen':
2357                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2358                         case 'currentmonthabbrev':
2359                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2360                         case 'currentday':
2361                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2362                         case 'currentday2':
2363                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2364                         case 'localmonth':
2365                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2366                         case 'localmonthname':
2367                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2368                         case 'localmonthnamegen':
2369                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2370                         case 'localmonthabbrev':
2371                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2372                         case 'localday':
2373                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2374                         case 'localday2':
2375                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2376                         case 'pagename':
2377                                 return wfEscapeWikiText( $this->mTitle->getText() );
2378                         case 'pagenamee':
2379                                 return $this->mTitle->getPartialURL();
2380                         case 'fullpagename':
2381                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2382                         case 'fullpagenamee':
2383                                 return $this->mTitle->getPrefixedURL();
2384                         case 'subpagename':
2385                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2386                         case 'subpagenamee':
2387                                 return $this->mTitle->getSubpageUrlForm();
2388                         case 'basepagename':
2389                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2390                         case 'basepagenamee':
2391                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2392                         case 'talkpagename':
2393                                 if( $this->mTitle->canTalk() ) {
2394                                         $talkPage = $this->mTitle->getTalkPage();
2395                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2396                                 } else {
2397                                         return '';
2398                                 }
2399                         case 'talkpagenamee':
2400                                 if( $this->mTitle->canTalk() ) {
2401                                         $talkPage = $this->mTitle->getTalkPage();
2402                                         return $talkPage->getPrefixedUrl();
2403                                 } else {
2404                                         return '';
2405                                 }
2406                         case 'subjectpagename':
2407                                 $subjPage = $this->mTitle->getSubjectPage();
2408                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2409                         case 'subjectpagenamee':
2410                                 $subjPage = $this->mTitle->getSubjectPage();
2411                                 return $subjPage->getPrefixedUrl();
2412                         case 'revisionid':
2413                                 // Let the edit saving system know we should parse the page
2414                                 // *after* a revision ID has been assigned.
2415                                 $this->mOutput->setFlag( 'vary-revision' );
2416                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2417                                 return $this->mRevisionId;
2418                         case 'revisionday':
2419                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2420                         case 'revisionday2':
2421                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2422                         case 'revisionmonth':
2423                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2424                         case 'revisionyear':
2425                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2426                         case 'revisiontimestamp':
2427                                 return $this->getRevisionTimestamp();
2428                         case 'namespace':
2429                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2430                         case 'namespacee':
2431                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2432                         case 'talkspace':
2433                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2434                         case 'talkspacee':
2435                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2436                         case 'subjectspace':
2437                                 return $this->mTitle->getSubjectNsText();
2438                         case 'subjectspacee':
2439                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2440                         case 'currentdayname':
2441                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2442                         case 'currentyear':
2443                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2444                         case 'currenttime':
2445                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2446                         case 'currenthour':
2447                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2448                         case 'currentweek':
2449                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2450                                 // int to remove the padding
2451                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2452                         case 'currentdow':
2453                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2454                         case 'localdayname':
2455                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2456                         case 'localyear':
2457                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2458                         case 'localtime':
2459                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2460                         case 'localhour':
2461                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2462                         case 'localweek':
2463                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2464                                 // int to remove the padding
2465                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2466                         case 'localdow':
2467                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2468                         case 'numberofarticles':
2469                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2470                         case 'numberoffiles':
2471                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2472                         case 'numberofusers':
2473                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2474                         case 'numberofpages':
2475                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2476                         case 'numberofadmins':
2477                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2478                         case 'numberofedits':
2479                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2480                         case 'currenttimestamp':
2481                                 return $varCache[$index] = wfTimestampNow();
2482                         case 'localtimestamp':
2483                                 return $varCache[$index] = $localTimestamp;
2484                         case 'currentversion':
2485                                 return $varCache[$index] = SpecialVersion::getVersion();
2486                         case 'sitename':
2487                                 return $wgSitename;
2488                         case 'server':
2489                                 return $wgServer;
2490                         case 'servername':
2491                                 return $wgServerName;
2492                         case 'scriptpath':
2493                                 return $wgScriptPath;
2494                         case 'directionmark':
2495                                 return $wgContLang->getDirMark();
2496                         case 'contentlanguage':
2497                                 global $wgContLanguageCode;
2498                                 return $wgContLanguageCode;
2499                         default:
2500                                 $ret = null;
2501                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2502                                         return $ret;
2503                                 else
2504                                         return null;
2505                 }
2506         }
2507
2508         /**
2509          * initialise the magic variables (like CURRENTMONTHNAME)
2510          *
2511          * @private
2512          */
2513         function initialiseVariables() {
2514                 $fname = 'Parser::initialiseVariables';
2515                 wfProfileIn( $fname );
2516                 $variableIDs = MagicWord::getVariableIDs();
2517
2518                 $this->mVariables = new MagicWordArray( $variableIDs );
2519                 wfProfileOut( $fname );
2520         }
2521
2522         /**
2523          * Parse any parentheses in format ((title|part|part)} and return the document tree
2524          * This is the ghost of replace_variables().
2525          *
2526          * @param string $text The text to parse
2527          * @private
2528          */
2529         function preprocessToDom ( $text ) {
2530                 wfProfileIn( __METHOD__ );
2531                 wfProfileIn( __METHOD__.'-makexml' );
2532
2533                 static $msgRules, $normalRules;
2534                 if ( !$msgRules ) {
2535                         $msgRules = array(
2536                                 '{' => array(
2537                                         'end' => '}',
2538                                         'names' => array(
2539                                                 2 => 'template',
2540                                         ),
2541                                         'min' => 2,
2542                                         'max' => 2,
2543                                 ),
2544                                 '[' => array(
2545                                         'end' => ']',
2546                                         'names' => array( 2 => null ),
2547                                         'min' => 2,
2548                                         'max' => 2,
2549                                 )
2550                         );
2551                         $normalRules = array(
2552                                 '{' => array(
2553                                         'end' => '}',
2554                                         'names' => array(
2555                                                 2 => 'template',
2556                                                 3 => 'tplarg',
2557                                         ),
2558                                         'min' => 2,
2559                                         'max' => 3,
2560                                 ),
2561                                 '[' => array(
2562                                         'end' => ']',
2563                                         'names' => array( 2 => null ),
2564                                         'min' => 2,
2565                                         'max' => 2,
2566                                 )
2567                         );
2568                 }
2569                 if ( $this->ot['msg'] ) {
2570                         $rules = $msgRules;
2571                 } else {
2572                         $rules = $normalRules;
2573                 }
2574
2575                 $extElements = implode( '|', $this->getStripList() );
2576                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2577                 $extElementsRegex = "/($extElements)(?:\s|\/>|>)|(!--)/iA";
2578
2579                 $stack = array();      # Stack of unclosed parentheses
2580                 $stackIndex = -1;      # Stack read pointer
2581
2582                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2583
2584                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2585                 $topAccum = '<root>';      # Top level text accumulator
2586                 $accum =& $topAccum;            # Current text accumulator
2587                 $findEquals = false;            # True to find equals signs in arguments
2588                 $findHeading = false;           # True to look at LF characters for possible headings
2589                 $findPipe = false;              # True to take notice of pipe characters
2590                 $headingIndex = 1;
2591                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2592
2593                 while ( $i < strlen( $text ) ) {
2594                         if ( $i == -1 ) {
2595                                 $found = 'line-start';
2596                                 $curChar = '';
2597                         } else {
2598                                 # Find next opening brace, closing brace or pipe
2599                                 $search = $searchBase;
2600                                 if ( $stackIndex == -1 ) {
2601                                         $currentClosing = '';
2602                                         // Look for headings only at the top stack level
2603                                         // Among other things, this resolves the ambiguity between =
2604                                         // for headings and = for template arguments
2605                                         $search .= "\n";
2606                                 } else {
2607                                         $currentClosing = $stack[$stackIndex]['close'];
2608                                         $search .= $currentClosing;
2609                                 }
2610                                 if ( $findPipe ) {
2611                                         $search .= '|';
2612                                 }
2613                                 if ( $findEquals ) {
2614                                         $search .= '=';
2615                                 }
2616                                 $rule = null;
2617                                 # Output literal section, advance input counter
2618                                 $literalLength = strcspn( $text, $search, $i );
2619                                 if ( $literalLength > 0 ) {
2620                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2621                                         $i += $literalLength;
2622                                 }
2623                                 if ( $i >= strlen( $text ) ) {
2624                                         if ( $currentClosing == "\n" ) {
2625                                                 // Do a past-the-end run to finish off the heading
2626                                                 $curChar = '';
2627                                                 $found = 'line-end';
2628                                         } else {
2629                                                 # All done
2630                                                 break;
2631                                         }
2632                                 } else {
2633                                         $curChar = $text[$i];
2634                                         if ( $curChar == '|' ) {
2635                                                 $found = 'pipe';
2636                                         } elseif ( $curChar == '=' ) {
2637                                                 $found = 'equals';
2638                                         } elseif ( $curChar == '<' ) {
2639                                                 $found = 'angle';
2640                                         } elseif ( $curChar == "\n" ) {
2641                                                 if ( $stackIndex == -1 ) {
2642                                                         $found = 'line-start';
2643                                                 } else {
2644                                                         $found = 'line-end';
2645                                                 }
2646                                         } elseif ( $curChar == $currentClosing ) {
2647                                                 $found = 'close';
2648                                         } elseif ( isset( $rules[$curChar] ) ) {
2649                                                 $found = 'open';
2650                                                 $rule = $rules[$curChar];
2651                                         } else {
2652                                                 # Some versions of PHP have a strcspn which stops on null characters
2653                                                 # Ignore and continue
2654                                                 ++$i;
2655                                                 continue;
2656                                         }
2657                                 }
2658                         }
2659
2660                         if ( $found == 'angle' ) {
2661                                 $matches = false;
2662                                 // Determine element name
2663                                 if ( !preg_match( $extElementsRegex, $text, $matches, 0, $i + 1 ) ) {
2664                                         // Element name missing or not listed
2665                                         $accum .= '&lt;';
2666                                         ++$i;
2667                                         continue;
2668                                 }
2669                                 // Handle comments
2670                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2671                                         // HTML comment, scan to end
2672                                         $endpos = strpos( $text, '-->', $i + 4 );
2673                                         if ( $endpos === false ) {
2674                                                 // Unclosed comment in input, runs to end
2675                                                 $accum .= htmlspecialchars( substr( $text, $i ) );
2676                                                 if ( $this->ot['html'] ) {
2677                                                         // Close it so later stripping can remove it
2678                                                         $accum .= htmlspecialchars( '-->' );
2679                                                 }
2680                                                 $i = strlen( $text );
2681                                                 continue;
2682                                         }
2683                                         $accum .= htmlspecialchars( substr( $text, $i, $endpos - $i + 3 ) );
2684                                         #$inner = substr( $text, $i + 4, $endpos - $i - 4 );
2685                                         #$accum .= '<ext><name>!--</name><inner>' . htmlspecialchars( $inner ) . '</inner></ext>';
2686                                         $i = $endpos + 3;
2687                                         continue;
2688                                 }
2689                                 $name = $matches[1];
2690                                 $attrStart = $i + strlen( $name ) + 1;
2691
2692                                 // Find end of tag
2693                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2694                                 if ( $tagEndPos === false ) {
2695                                         // Infinite backtrack
2696                                         // Disable tag search to prevent worst-case O(N^2) performance
2697                                         $noMoreGT = true;
2698                                         $accum .= '&lt;';
2699                                         ++$i;
2700                                         continue;
2701                                 }
2702                                 if ( $text[$tagEndPos-1] == '/' ) {
2703                                         $attrEnd = $tagEndPos - 1;
2704                                         $inner = null;
2705                                         $i = $tagEndPos + 1;
2706                                         $close = '';
2707                                 } else {
2708                                         $attrEnd = $tagEndPos;
2709                                         // Find closing tag
2710                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2711                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2712                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2713                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2714                                         } else {
2715                                                 // No end tag -- let it run out to the end of the text.
2716                                                 $inner = substr( $text, $tagEndPos + 1 );
2717                                                 $i = strlen( $text );
2718                                                 $close = '';
2719                                         }
2720                                 }
2721                                 $accum .= '<ext>';
2722                                 if ( $attrEnd <= $attrStart ) {
2723                                         $attr = '';
2724                                 } else {
2725                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2726                                 }
2727                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2728                                         // Note that the attr element contains the whitespace between name and attribute,
2729                                         // this is necessary for precise reconstruction during pre-save transform.
2730                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2731                                 if ( $inner !== null ) {
2732                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2733                                 }
2734                                 $accum .= $close . '</ext>';
2735                         }
2736
2737                         elseif ( $found == 'line-start' ) {
2738                                 // Is this the start of a heading?
2739                                 // Line break belongs before the heading element in any case
2740                                 $accum .= $curChar;
2741                                 $i++;
2742
2743                                 $count = strspn( $text, '=', $i, 6 );
2744                                 if ( $count > 0 ) {
2745                                         $piece = array(
2746                                                 'open' => "\n",
2747                                                 'close' => "\n",
2748                                                 'parts' => array( str_repeat( '=', $count ) ),
2749                                                 'count' => $count );
2750                                         $stack[++$stackIndex] = $piece;
2751                                         $i += $count;
2752                                         $accum =& $stack[$stackIndex]['parts'][0];
2753                                         $findPipe = false;
2754                                 }
2755                         }
2756
2757                         elseif ( $found == 'line-end' ) {
2758                                 $piece = $stack[$stackIndex];
2759                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2760                                 assert( $piece['open'] == "\n" );
2761                                 assert( $stackIndex == 0 );
2762                                 // Search back through the accumulator to see if it has a proper close
2763                                 // No efficient way to do this in PHP AFAICT: strrev, PCRE search with $ anchor
2764                                 // and rtrim are all O(N) in total size. Optimal would be O(N) in trailing
2765                                 // whitespace size only.
2766                                 $m = false;
2767                                 $count = $piece['count'];
2768                                 if ( preg_match( "/(={{$count}})\s*$/", $accum, $m, 0, $count ) ) {
2769                                         // Found match, output <h>
2770                                         $count = min( strlen( $m[1] ), $count );
2771                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2772                                         $headingIndex++;
2773                                 } else {
2774                                         // No match, no <h>, just pass down the inner text
2775                                         $element = $accum;
2776                                 }
2777                                 // Unwind the stack
2778                                 // Headings can only occur on the top level, so this is a bit simpler than the
2779                                 // generic stack unwind operation in the close case
2780                                 unset( $stack[$stackIndex--] );
2781                                 $accum =& $topAccum;
2782                                 $findEquals = false;
2783                                 $findPipe = false;
2784
2785                                 // Append the result to the enclosing accumulator
2786                                 $accum .= $element;
2787                                 // Note that we do NOT increment the input pointer.
2788                                 // This is because the closing linebreak could be the opening linebreak of
2789                                 // another heading. Infinite loops are avoided because the next iteration MUST
2790                                 // hit the heading open case above, which unconditionally increments the
2791                                 // input pointer.
2792                         }
2793
2794                         elseif ( $found == 'open' ) {
2795                                 # count opening brace characters
2796                                 $count = strspn( $text, $curChar, $i );
2797
2798                                 # we need to add to stack only if opening brace count is enough for one of the rules
2799                                 if ( $count >= $rule['min'] ) {
2800                                         # Add it to the stack
2801                                         $piece = array(
2802                                                 'open' => $curChar,
2803                                                 'close' => $rule['end'],
2804                                                 'count' => $count,
2805                                                 'parts' => array( '' ),
2806                                                 'eqpos' => array(),
2807                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2808                                         );
2809
2810                                         $stackIndex ++;
2811                                         $stack[$stackIndex] = $piece;
2812                                         $accum =& $stack[$stackIndex]['parts'][0];
2813                                         $findEquals = false;
2814                                         $findPipe = true;
2815                                 } else {
2816                                         # Add literal brace(s)
2817                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2818                                 }
2819                                 $i += $count;
2820                         }
2821
2822                         elseif ( $found == 'close' ) {
2823                                 $piece = $stack[$stackIndex];
2824                                 # lets check if there are enough characters for closing brace
2825                                 $maxCount = $piece['count'];
2826                                 $count = strspn( $text, $curChar, $i, $maxCount );
2827
2828                                 # check for maximum matching characters (if there are 5 closing
2829                                 # characters, we will probably need only 3 - depending on the rules)
2830                                 $matchingCount = 0;
2831                                 $rule = $rules[$piece['open']];
2832                                 if ( $count > $rule['max'] ) {
2833                                         # The specified maximum exists in the callback array, unless the caller
2834                                         # has made an error
2835                                         $matchingCount = $rule['max'];
2836                                 } else {
2837                                         # Count is less than the maximum
2838                                         # Skip any gaps in the callback array to find the true largest match
2839                                         # Need to use array_key_exists not isset because the callback can be null
2840                                         $matchingCount = $count;
2841                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2842                                                 --$matchingCount;
2843                                         }
2844                                 }
2845
2846                                 if ($matchingCount <= 0) {
2847                                         # No matching element found in callback array
2848                                         # Output a literal closing brace and continue
2849                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2850                                         $i += $count;
2851                                         continue;
2852                                 }
2853                                 $name = $rule['names'][$matchingCount];
2854                                 if ( $name === null ) {
2855                                         // No element, just literal text
2856                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2857                                                 implode( '|', $piece['parts'] ) .
2858                                                 str_repeat( $rule['end'], $matchingCount );
2859                                 } else {
2860                                         # Create XML element
2861                                         # Note: $parts is already XML, does not need to be encoded further
2862                                         $parts = $piece['parts'];
2863                                         $title = $parts[0];
2864                                         unset( $parts[0] );
2865
2866                                         # The invocation is at the start of the line if lineStart is set in
2867                                         # the stack, and all opening brackets are used up.
2868                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
2869                                                 $attr = ' lineStart="1"';
2870                                         } else {
2871                                                 $attr = '';
2872                                         }
2873
2874                                         $element = "<$name$attr>";
2875                                         $element .= "<title>$title</title>";
2876                                         $argIndex = 1;
2877                                         foreach ( $parts as $partIndex => $part ) {
2878                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
2879                                                         $eqpos = $piece['eqpos'][$partIndex];
2880                                                         list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
2881                                                         list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
2882                                                         $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
2883                                                 } else {
2884                                                         list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
2885                                                         $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
2886                                                         $argIndex++;
2887                                                 }
2888                                         }
2889                                         $element .= "</$name>";
2890                                 }
2891
2892                                 # Advance input pointer
2893                                 $i += $matchingCount;
2894
2895                                 # Unwind the stack
2896                                 unset( $stack[$stackIndex--] );
2897                                 if ( $stackIndex == -1 ) {
2898                                         $accum =& $topAccum;
2899                                         $findEquals = false;
2900                                         $findPipe = false;
2901                                 } else {
2902                                         $partCount = count( $stack[$stackIndex]['parts'] );
2903                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
2904                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
2905                                         $findEquals = $findPipe && $partCount > 1
2906                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
2907                                 }
2908
2909                                 # Re-add the old stack element if it still has unmatched opening characters remaining
2910                                 if ($matchingCount < $piece['count']) {
2911                                         $piece['parts'] = array( '' );
2912                                         $piece['count'] -= $matchingCount;
2913                                         $piece['eqpos'] = array();
2914                                         # do we still qualify for any callback with remaining count?
2915                                         $names = $rules[$piece['open']]['names'];
2916                                         $skippedBraces = 0;
2917                                         $enclosingAccum =& $accum;
2918                                         while ( $piece['count'] ) {
2919                                                 if ( array_key_exists( $piece['count'], $names ) ) {
2920                                                         $stackIndex++;
2921                                                         $stack[$stackIndex] = $piece;
2922                                                         $accum =& $stack[$stackIndex]['parts'][0];
2923                                                         $findEquals = true;
2924                                                         $findPipe = true;
2925                                                         break;
2926                                                 }
2927                                                 --$piece['count'];
2928                                                 $skippedBraces ++;
2929                                         }
2930                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
2931                                 }
2932
2933                                 # Add XML element to the enclosing accumulator
2934                                 $accum .= $element;
2935                         }
2936
2937                         elseif ( $found == 'pipe' ) {
2938                                 $stack[$stackIndex]['parts'][] = '';
2939                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2940                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
2941                                 $findEquals = true;
2942                                 ++$i;
2943                         }
2944
2945                         elseif ( $found == 'equals' ) {
2946                                 $findEquals = false;
2947                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2948                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
2949                                 $accum .= '=';
2950                                 ++$i;
2951                         }
2952                 }
2953
2954                 # Output any remaining unclosed brackets
2955                 foreach ( $stack as $piece ) {
2956                         if ( $piece['open'] == "\n" ) {
2957                                 $topAccum .= $piece['parts'][0];
2958                         } else {
2959                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
2960                         }
2961                 }
2962                 $topAccum .= '</root>';
2963
2964                 wfProfileOut( __METHOD__.'-makexml' );
2965                 wfProfileIn( __METHOD__.'-loadXML' );
2966                 $dom = new DOMDocument;
2967                 if ( !$dom->loadXML( $topAccum ) ) {
2968                         throw new MWException( __METHOD__.' generated invalid XML' );
2969                 }
2970                 wfProfileOut( __METHOD__.'-loadXML' );
2971                 wfProfileOut( __METHOD__ );
2972                 return $dom;
2973         }
2974
2975         /*
2976          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2977          */
2978         public static function splitWhitespace( $s ) {
2979                 $ltrimmed = ltrim( $s );
2980                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2981                 $trimmed = rtrim( $ltrimmed );
2982                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2983                 if ( $diff > 0 ) {
2984                         $w2 = substr( $ltrimmed, -$diff );
2985                 } else {
2986                         $w2 = '';
2987                 }
2988                 return array( $w1, $trimmed, $w2 );
2989         }
2990
2991         /**
2992          * Replace magic variables, templates, and template arguments
2993          * with the appropriate text. Templates are substituted recursively,
2994          * taking care to avoid infinite loops.
2995          *
2996          * Note that the substitution depends on value of $mOutputType:
2997          *  OT_WIKI: only {{subst:}} templates
2998          *  OT_MSG: only magic variables
2999          *  OT_HTML: all templates and magic variables
3000          *
3001          * @param string $tex The text to transform
3002          * @param PPFrame $frame Object describing the arguments passed to the template
3003          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3004          * @private
3005          */
3006         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3007                 # Prevent too big inclusions
3008                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3009                         return $text;
3010                 }
3011
3012                 $fname = __METHOD__;
3013                 wfProfileIn( $fname );
3014
3015                 if ( $frame === false ) {
3016                         $frame = new PPFrame( $this );
3017                 } elseif ( !( $frame instanceof PPFrame ) ) {
3018                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3019                 }
3020
3021                 $dom = $this->preprocessToDom( $text );
3022                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3023                 $text = $frame->expand( $dom, 0, $flags );
3024
3025                 wfProfileOut( $fname );
3026                 return $text;
3027         }
3028
3029         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3030         static function createAssocArgs( $args ) {
3031                 $assocArgs = array();
3032                 $index = 1;
3033                 foreach( $args as $arg ) {
3034                         $eqpos = strpos( $arg, '=' );
3035                         if ( $eqpos === false ) {
3036                                 $assocArgs[$index++] = $arg;
3037                         } else {
3038                                 $name = trim( substr( $arg, 0, $eqpos ) );
3039                                 $value = trim( substr( $arg, $eqpos+1 ) );
3040                                 if ( $value === false ) {
3041                                         $value = '';
3042                                 }
3043                                 if ( $name !== false ) {
3044                                         $assocArgs[$name] = $value;
3045                                 }
3046                         }
3047                 }
3048
3049                 return $assocArgs;
3050         }
3051
3052         /**
3053          * Return the text of a template, after recursively
3054          * replacing any variables or templates within the template.
3055          *
3056          * @param array $piece The parts of the template
3057          *  $piece['text']: matched text
3058          *  $piece['title']: the title, i.e. the part before the |
3059          *  $piece['parts']: the parameter array
3060          * @param PPFrame The current frame, contains template arguments
3061          * @return string the text of the template
3062          * @private
3063          */
3064         function braceSubstitution( $piece, $frame ) {
3065                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3066                 $fname = __METHOD__;
3067                 wfProfileIn( $fname );
3068                 wfProfileIn( __METHOD__.'-setup' );
3069
3070                 # Flags
3071                 $found = false;             # $text has been filled
3072                 $nowiki = false;            # wiki markup in $text should be escaped
3073                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
3074                 $noargs = false;            # Don't replace triple-brace arguments in $text
3075                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3076                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3077                 $isDOM = false;             # $text is a DOM node needing expansion
3078
3079                 # Title object, where $text came from
3080                 $title = NULL;
3081
3082                 # $part1 is the bit before the first |, and must contain only title characters.
3083                 # Various prefixes will be stripped from it later.
3084                 $titleWithSpaces = $frame->expand( $piece['title'] );
3085                 $part1 = trim( $titleWithSpaces );
3086                 $titleText = false;
3087
3088                 # Original title text preserved for various purposes
3089                 $originalTitle = $part1;
3090
3091                 # $args is a list of argument nodes, starting from index 0, not including $part1
3092                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3093                 wfProfileOut( __METHOD__.'-setup' );
3094
3095                 # SUBST
3096                 wfProfileIn( __METHOD__.'-modifiers' );
3097                 if ( !$found ) {
3098                         $mwSubst =& MagicWord::get( 'subst' );
3099                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3100                                 # One of two possibilities is true:
3101                                 # 1) Found SUBST but not in the PST phase
3102                                 # 2) Didn't find SUBST and in the PST phase
3103                                 # In either case, return without further processing
3104                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3105                                 $found = true;
3106                                 $noparse = true;
3107                                 $noargs = true;
3108                         }
3109                 }
3110
3111                 # Variables
3112                 if ( !$found && $args->length == 0 ) {
3113                         $id = $this->mVariables->matchStartToEnd( $part1 );
3114                         if ( $id !== false ) {
3115                                 $text = $this->getVariableValue( $id );
3116                                 $this->mOutput->mContainsOldMagic = true;
3117                                 $found = true;
3118                                 $noparse = true;
3119                                 $noargs = true;
3120                         }
3121                 }
3122
3123                 # MSG, MSGNW and RAW
3124                 if ( !$found ) {
3125                         # Check for MSGNW:
3126                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3127                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3128                                 $nowiki = true;
3129                         } else {
3130                                 # Remove obsolete MSG:
3131                                 $mwMsg =& MagicWord::get( 'msg' );
3132                                 $mwMsg->matchStartAndRemove( $part1 );
3133                         }
3134
3135                         # Check for RAW:
3136                         $mwRaw =& MagicWord::get( 'raw' );
3137                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3138                                 $forceRawInterwiki = true;
3139                         }
3140                 }
3141                 wfProfileOut( __METHOD__.'-modifiers' );
3142
3143                 # Save path level before recursing into functions & templates.
3144                 $lastPathLevel = $this->mTemplatePath;
3145
3146                 # Parser functions
3147                 if ( !$found ) {
3148                         wfProfileIn( __METHOD__ . '-pfunc' );
3149
3150                         $colonPos = strpos( $part1, ':' );
3151                         if ( $colonPos !== false ) {
3152                                 # Case sensitive functions
3153                                 $function = substr( $part1, 0, $colonPos );
3154                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3155                                         $function = $this->mFunctionSynonyms[1][$function];
3156                                 } else {
3157                                         # Case insensitive functions
3158                                         $function = strtolower( $function );
3159                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3160                                                 $function = $this->mFunctionSynonyms[0][$function];
3161                                         } else {
3162                                                 $function = false;
3163                                         }
3164                                 }
3165                                 if ( $function ) {
3166                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3167                                         $initialArgs = array( &$this );
3168                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3169                                         if ( $flags & SFH_OBJECT_ARGS ) {
3170                                                 # Add a frame parameter, and pass the arguments as an array
3171                                                 $allArgs = $initialArgs;
3172                                                 $allArgs[] = $frame;
3173                                                 foreach ( $args as $arg ) {
3174                                                         $funcArgs[] = $arg;
3175                                                 }
3176                                                 $allArgs[] = $funcArgs;
3177                                         } else {
3178                                                 # Convert arguments to plain text
3179                                                 foreach ( $args as $arg ) {
3180                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3181                                                 }
3182                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3183                                         }
3184
3185                                         $result = call_user_func_array( $callback, $allArgs );
3186                                         $found = true;
3187
3188                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3189                                         $noargs = true;
3190                                         $noparse = true;
3191
3192                                         if ( is_array( $result ) ) {
3193                                                 if ( isset( $result[0] ) ) {
3194                                                         $text = $result[0];
3195                                                         unset( $result[0] );
3196                                                 }
3197
3198                                                 // Extract flags into the local scope
3199                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3200                                                 extract( $result );
3201                                         } else {
3202                                                 $text = $result;
3203                                         }
3204                                 }
3205                         }
3206                         wfProfileOut( __METHOD__ . '-pfunc' );
3207                 }
3208
3209                 # Finish mangling title and then check for loops.
3210                 # Set $title to a Title object and $titleText to the PDBK
3211                 if ( !$found ) {
3212                         $ns = NS_TEMPLATE;
3213                         # Split the title into page and subpage
3214                         $subpage = '';
3215                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3216                         if ($subpage !== '') {
3217                                 $ns = $this->mTitle->getNamespace();
3218                         }
3219                         $title = Title::newFromText( $part1, $ns );
3220                         if ( $title ) {
3221                                 $titleText = $title->getPrefixedText();
3222                                 # Check for language variants if the template is not found
3223                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3224                                         $wgContLang->findVariantLink($part1, $title);
3225                                 }
3226                                 # Do infinite loop check
3227                                 if ( isset( $this->mTemplatePath[$titleText] ) ) {
3228                                         $noparse = true;
3229                                         $noargs = true;
3230                                         $found = true;
3231                                         $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
3232                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3233                                 }
3234                         }
3235                 }
3236
3237                 # Load from database
3238                 if ( !$found && $title ) {
3239                         wfProfileIn( __METHOD__ . '-loadtpl' );
3240                         if ( !$title->isExternal() ) {
3241                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3242                                         $text = SpecialPage::capturePath( $title );
3243                                         if ( is_string( $text ) ) {
3244                                                 $found = true;
3245                                                 $noparse = true;
3246                                                 $noargs = true;
3247                                                 $isHTML = true;
3248                                                 $this->disableCache();
3249                                         }
3250                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3251                                         $found = false; //access denied
3252                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3253                                 } else {
3254                                         list( $text, $title ) = $this->getTemplateDom( $title );
3255                                         if ( $text !== false ) {
3256                                                 $found = true;
3257                                                 $isDOM = true;
3258                                         }
3259                                 }
3260
3261                                 # If the title is valid but undisplayable, make a link to it
3262                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3263                                         $text = "[[:$titleText]]";
3264                                         $found = true;
3265                                 }
3266                         } elseif ( $title->isTrans() ) {
3267                                 // Interwiki transclusion
3268                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3269                                         $text = $this->interwikiTransclude( $title, 'render' );
3270                                         $isHTML = true;
3271                                         $noparse = true;
3272                                 } else {
3273                                         $text = $this->interwikiTransclude( $title, 'raw' );
3274                                 }
3275                                 $found = true;
3276                         }
3277                         wfProfileOut( __METHOD__ . '-loadtpl' );
3278                 }
3279
3280                 # Recursive parsing, escaping and link table handling
3281                 # Only for HTML output
3282                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3283                         if ( $isDOM ) {
3284                                 $text = $frame->expand( $text );
3285                         }
3286                         $text = wfEscapeWikiText( $text );
3287                 } elseif ( !$this->ot['msg'] && $found ) {
3288                         if ( $noargs ) {
3289                                 $newFrame = $frame->newChild();
3290                         } else {
3291                                 # Clean up argument array
3292                                 $newFrame = $frame->newChild( $args, $title );
3293                                 # Add a new element to the templace recursion path
3294                                 $this->mTemplatePath[$titleText] = 1;
3295                         }
3296
3297                         if ( !$noparse ) {
3298                                 if ( $isDOM ) {
3299                                         if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
3300                                                 # Expansion is eligible for the empty-frame cache
3301                                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3302                                                         $text = $this->mTplExpandCache[$titleText];
3303                                                 } else {
3304                                                         $text = $newFrame->expand( $text );
3305                                                         $this->mTplExpandCache[$titleText] = $text;
3306                                                 }
3307                                         } else {
3308                                                 $text = $newFrame->expand( $text );
3309                                         }
3310                                 } else {
3311                                         $text = $this->replaceVariables( $text, $newFrame );
3312                                 }
3313
3314                                 # strip woz 'ere 2004-07
3315
3316                                 # Bug 529: if the template begins with a table or block-level
3317                                 # element, it should be treated as beginning a new line.
3318                                 # This behaviour is somewhat controversial.
3319                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3320                                         $text = "\n" . $text;
3321                                 }
3322                         } elseif ( !$noargs ) {
3323                                 # $noparse and !$noargs
3324                                 # Just replace the arguments, not any double-brace items
3325                                 # This is used for rendered interwiki transclusion
3326                                 if ( $isDOM ) {
3327                                         $text = $newFrame->expand( $text, 0, PPFrame::NO_TEMPLATES );
3328                                 } else {
3329                                         $text = $this->replaceVariables( $text, $newFrame, true );
3330                                 }
3331                         } elseif ( $isDOM ) {
3332                                 $text = $frame->expand( $text );
3333                         }
3334                 } elseif ( $isDOM ) {
3335                         $text = $frame->expand( $text, 0, PPFrame::NO_TEMPLATES | PPFrame::NO_ARGS );
3336                 }
3337
3338                 # Prune lower levels off the recursion check path
3339                 $this->mTemplatePath = $lastPathLevel;
3340
3341                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3342                         # Error, oversize inclusion
3343                         $text = "[[$originalTitle]]" .
3344                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3345                         $noparse = true;
3346                         $noargs = true;
3347                 }
3348
3349                 if ( !$found ) {
3350                         wfProfileOut( $fname );
3351                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3352                 } else {
3353                         wfProfileIn( __METHOD__ . '-placeholders' );
3354                         if ( $isHTML ) {
3355                                 # Replace raw HTML by a placeholder
3356                                 # Add a blank line preceding, to prevent it from mucking up
3357                                 # immediately preceding headings
3358                                 $text = "\n\n" . $this->insertStripItem( $text );
3359                         }
3360                         wfProfileOut( __METHOD__ . '-placeholders' );
3361                 }
3362
3363                 # Prune lower levels off the recursion check path
3364                 $this->mTemplatePath = $lastPathLevel;
3365
3366                 if ( !$found ) {
3367                         wfProfileOut( $fname );
3368                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3369                 } else {
3370                         wfProfileOut( $fname );
3371                         return $text;
3372                 }
3373         }
3374
3375         /**
3376          * Get the semi-parsed DOM representation of a template with a given title,
3377          * and its redirect destination title. Cached.
3378          */
3379         function getTemplateDom( $title ) {
3380                 $cacheTitle = $title;
3381                 $titleText = $title->getPrefixedDBkey();
3382
3383                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3384                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3385                         $title = Title::makeTitle( $ns, $dbk );
3386                         $titleText = $title->getPrefixedDBkey();
3387                 }
3388                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3389                         return array( $this->mTplDomCache[$titleText], $title );
3390                 }
3391
3392                 // Cache miss, go to the database
3393                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3394
3395                 if ( $text === false ) {
3396                         $this->mTplDomCache[$titleText] = false;
3397                         return array( false, $title );
3398                 }
3399
3400                 # If there are any <onlyinclude> tags, only include them
3401                 if ( !$this->ot['msg'] ) {
3402                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3403                                 $replacer = new OnlyIncludeReplacer;
3404                                 StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3405                                         array( &$replacer, 'replace' ), $text );
3406                                 $text = $replacer->output;
3407                         }
3408                         # Remove <noinclude> sections and <includeonly> tags
3409                         $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3410                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3411                 }
3412
3413                 $dom = $this->preprocessToDom( $text );
3414
3415                 $this->mTplDomCache[ $titleText ] = $dom;
3416
3417                 if (! $title->equals($cacheTitle)) {
3418                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3419                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3420                 }
3421
3422                 return array( $dom, $title );
3423         }
3424
3425         /**
3426          * Fetch the unparsed text of a template and register a reference to it.
3427          */
3428         function fetchTemplateAndTitle( $title ) {
3429                 $templateCb = $this->mOptions->getTemplateCallback();
3430                 $stuff = call_user_func( $templateCb, $title );
3431                 $text = $stuff['text'];
3432                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3433                 if ( isset( $stuff['deps'] ) ) {
3434                         foreach ( $stuff['deps'] as $dep ) {
3435                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3436                         }
3437                 }
3438                 return array($text,$finalTitle);
3439         }
3440
3441         function fetchTemplate( $title ) {
3442                 $rv = $this->fetchTemplateAndTitle($title);
3443                 return $rv[0];
3444         }
3445
3446         /**
3447          * Static function to get a template
3448          * Can be overridden via ParserOptions::setTemplateCallback().
3449          */
3450         static function statelessFetchTemplate( $title ) {
3451                 $text = $skip = false;
3452                 $finalTitle = $title;
3453                 $deps = array();
3454
3455                 // Loop to fetch the article, with up to 1 redirect
3456                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3457                         # Give extensions a chance to select the revision instead
3458                         $id = false; // Assume current
3459                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3460
3461                         if( $skip ) {
3462                                 $text = false;
3463                                 $deps[] = array(
3464                                         'title' => $title,
3465                                         'page_id' => $title->getArticleID(),
3466                                         'rev_id' => null );
3467                                 break;
3468                         }
3469                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3470                         $rev_id = $rev ? $rev->getId() : 0;
3471
3472                         $deps[] = array(
3473                                 'title' => $title,
3474                                 'page_id' => $title->getArticleID(),
3475                                 'rev_id' => $rev_id );
3476
3477                         if( $rev ) {
3478                                 $text = $rev->getText();
3479                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3480                                 global $wgLang;
3481                                 $message = $wgLang->lcfirst( $title->getText() );
3482                                 $text = wfMsgForContentNoTrans( $message );
3483                                 if( wfEmptyMsg( $message, $text ) ) {
3484                                         $text = false;
3485                                         break;
3486                                 }
3487                         } else {
3488                                 break;
3489                         }
3490                         if ( $text === false ) {
3491                                 break;
3492                         }
3493                         // Redirect?
3494                         $finalTitle = $title;
3495                         $title = Title::newFromRedirect( $text );
3496                 }
3497                 return array(
3498                         'text' => $text,
3499                         'finalTitle' => $finalTitle,
3500                         'deps' => $deps );
3501         }
3502
3503         /**
3504          * Transclude an interwiki link.
3505          */
3506         function interwikiTransclude( $title, $action ) {
3507                 global $wgEnableScaryTranscluding;
3508
3509                 if (!$wgEnableScaryTranscluding)
3510                         return wfMsg('scarytranscludedisabled');
3511
3512                 $url = $title->getFullUrl( "action=$action" );
3513
3514                 if (strlen($url) > 255)
3515                         return wfMsg('scarytranscludetoolong');
3516                 return $this->fetchScaryTemplateMaybeFromCache($url);
3517         }
3518
3519         function fetchScaryTemplateMaybeFromCache($url) {
3520                 global $wgTranscludeCacheExpiry;
3521                 $dbr = wfGetDB(DB_SLAVE);
3522                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3523                                 array('tc_url' => $url));
3524                 if ($obj) {
3525                         $time = $obj->tc_time;
3526                         $text = $obj->tc_contents;
3527                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3528                                 return $text;
3529                         }
3530                 }
3531
3532                 $text = Http::get($url);
3533                 if (!$text)
3534                         return wfMsg('scarytranscludefailed', $url);
3535
3536                 $dbw = wfGetDB(DB_MASTER);
3537                 $dbw->replace('transcache', array('tc_url'), array(
3538                         'tc_url' => $url,
3539                         'tc_time' => time(),
3540                         'tc_contents' => $text));
3541                 return $text;
3542         }
3543
3544
3545         /**
3546          * Triple brace replacement -- used for template arguments
3547          * @private
3548          */
3549         function argSubstitution( $piece, $frame ) {
3550                 wfProfileIn( __METHOD__ );
3551
3552                 $text = false;
3553                 $error = false;
3554                 $parts = $piece['parts'];
3555                 $argWithSpaces = $frame->expand( $piece['title'] );
3556                 $arg = trim( $argWithSpaces );
3557
3558                 if ( isset( $frame->args[$arg] ) ) {
3559                         $text = $frame->parent->expand( $frame->args[$arg] );
3560                 } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3561                         $text = $frame->expand( $parts->item( 0 ) );
3562                 }
3563                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3564                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3565                 }
3566
3567                 if ( $text === false ) {
3568                         $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
3569                 }
3570                 if ( $error !== false ) {
3571                         $text .= $error;
3572                 }
3573
3574                 wfProfileOut( __METHOD__ );
3575                 return $text;
3576         }
3577
3578         /**
3579          * Return the text to be used for a given extension tag.
3580          * This is the ghost of strip().
3581          *
3582          * @param array $params Associative array of parameters:
3583          *     name       DOMNode for the tag name
3584          *     attrText   DOMNode for unparsed text where tag attributes are thought to be
3585          *     inner      Contents of extension element
3586          *     noClose    Original text did not have a close tag
3587          * @param PPFrame $frame
3588          */
3589         function extensionSubstitution( $params, $frame ) {
3590                 global $wgRawHtml, $wgContLang;
3591                 static $n = 1;
3592
3593                 $name = $frame->expand( $params['name'] );
3594                 $attrText = is_null( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3595                 $content = is_null( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3596
3597                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3598
3599                 if ( $this->ot['html'] ) {
3600                         if ( $name == '!--' ) {
3601                                 return '';
3602                         }
3603                         $name = strtolower( $name );
3604
3605                         $params = Sanitizer::decodeTagAttributes( $attrText );
3606                         switch ( $name ) {
3607                                 case 'html':
3608                                         if( $wgRawHtml ) {
3609                                                 $output = $content;
3610                                                 break;
3611                                         } else {
3612                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3613                                         }
3614                                 case 'nowiki':
3615                                         $output = Xml::escapeTagsOnly( $content );
3616                                         break;
3617                                 case 'math':
3618                                         $output = $wgContLang->armourMath(
3619                                                 MathRenderer::renderMath( $content, $params ) );
3620                                         break;
3621                                 case 'gallery':
3622                                         $output = $this->renderImageGallery( $content, $params );
3623                                         break;
3624                                 default:
3625                                         if( isset( $this->mTagHooks[$name] ) ) {
3626                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3627                                                         array( $content, $params, $this ) );
3628                                         } else {
3629                                                 throw new MWException( "Invalid call hook $name" );
3630                                         }
3631                         }
3632                 } else {
3633                         if ( $name == '!--' ) {
3634                                 $output = '<!--' . $content . '-->';
3635                         } else {
3636                                 if ( $content === null ) {
3637                                         $output = "<$name$attrText/>";
3638                                 } else {
3639                                         $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3640                                         $output = "<$name$attrText>$content$close";
3641                                 }
3642                         }
3643                 }
3644
3645                 if ( $name == 'html' || $name == 'nowiki' ) {
3646                         $this->mStripState->nowiki->setPair( $marker, $output );
3647                 } else {
3648                         $this->mStripState->general->setPair( $marker, $output );
3649                 }
3650                 return $marker;
3651         }
3652
3653         /**
3654          * Increment an include size counter
3655          *
3656          * @param string $type The type of expansion
3657          * @param integer $size The size of the text
3658          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3659          */
3660         function incrementIncludeSize( $type, $size ) {
3661                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3662                         return false;
3663                 } else {
3664                         $this->mIncludeSizes[$type] += $size;
3665                         return true;
3666                 }
3667         }
3668
3669         /**
3670          * Detect __NOGALLERY__ magic word and set a placeholder
3671          */
3672         function stripNoGallery( &$text ) {
3673                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3674                 # do not add TOC
3675                 $mw = MagicWord::get( 'nogallery' );
3676                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3677         }
3678
3679         /**
3680          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3681          * placeholder that will then be replaced by the real TOC in
3682          * ->formatHeadings, this works because at this points real
3683          * comments will have already been discarded by the sanitizer.
3684          *
3685          * Any additional __TOC__ magic words left over will be discarded
3686          * as there can only be one TOC on the page.
3687          */
3688         function stripToc( $text ) {
3689                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3690                 # do not add TOC
3691                 $mw = MagicWord::get( 'notoc' );
3692                 if( $mw->matchAndRemove( $text ) ) {
3693                         $this->mShowToc = false;
3694                 }
3695
3696                 $mw = MagicWord::get( 'toc' );
3697                 if( $mw->match( $text ) ) {
3698                         $this->mShowToc = true;
3699                         $this->mForceTocPosition = true;
3700
3701                         // Set a placeholder. At the end we'll fill it in with the TOC.
3702                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3703
3704                         // Only keep the first one.
3705                         $text = $mw->replace( '', $text );
3706                 }
3707                 return $text;
3708         }
3709
3710         /**
3711          * This function accomplishes several tasks:
3712          * 1) Auto-number headings if that option is enabled
3713          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3714          * 3) Add a Table of contents on the top for users who have enabled the option
3715          * 4) Auto-anchor headings
3716          *
3717          * It loops through all headlines, collects the necessary data, then splits up the
3718          * string and re-inserts the newly formatted headlines.
3719          *
3720          * @param string $text
3721          * @param boolean $isMain
3722          * @private
3723          */
3724         function formatHeadings( $text, $isMain=true ) {
3725                 global $wgMaxTocLevel, $wgContLang;
3726
3727                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3728                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3729                         $showEditLink = 0;
3730                 } else {
3731                         $showEditLink = $this->mOptions->getEditSection();
3732                 }
3733
3734                 # Inhibit editsection links if requested in the page
3735                 $esw =& MagicWord::get( 'noeditsection' );
3736                 if( $esw->matchAndRemove( $text ) ) {
3737                         $showEditLink = 0;
3738                 }
3739
3740                 # Get all headlines for numbering them and adding funky stuff like [edit]
3741                 # links - this is for later, but we need the number of headlines right now
3742                 $matches = array();
3743                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3744
3745                 # if there are fewer than 4 headlines in the article, do not show TOC
3746                 # unless it's been explicitly enabled.
3747                 $enoughToc = $this->mShowToc &&
3748                         (($numMatches >= 4) || $this->mForceTocPosition);
3749
3750                 # Allow user to stipulate that a page should have a "new section"
3751                 # link added via __NEWSECTIONLINK__
3752                 $mw =& MagicWord::get( 'newsectionlink' );
3753                 if( $mw->matchAndRemove( $text ) )
3754                         $this->mOutput->setNewSection( true );
3755
3756                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3757                 # override above conditions and always show TOC above first header
3758                 $mw =& MagicWord::get( 'forcetoc' );
3759                 if ($mw->matchAndRemove( $text ) ) {
3760                         $this->mShowToc = true;
3761                         $enoughToc = true;
3762                 }
3763
3764                 # We need this to perform operations on the HTML
3765                 $sk = $this->mOptions->getSkin();
3766
3767                 # headline counter
3768                 $headlineCount = 0;
3769                 $numVisible = 0;
3770
3771                 # Ugh .. the TOC should have neat indentation levels which can be
3772                 # passed to the skin functions. These are determined here
3773                 $toc = '';
3774                 $full = '';
3775                 $head = array();
3776                 $sublevelCount = array();
3777                 $levelCount = array();
3778                 $toclevel = 0;
3779                 $level = 0;
3780                 $prevlevel = 0;
3781                 $toclevel = 0;
3782                 $prevtoclevel = 0;
3783                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3784                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3785
3786                 foreach( $matches[3] as $headline ) {
3787                         $isTemplate = false;
3788                         $titleText = false;
3789                         $sectionIndex = false;
3790                         $numbering = '';
3791                         $markerMatches = array();
3792                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3793                                 $serial = $markerMatches[1];
3794                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3795                                 $isTemplate = ($titleText != $baseTitleText);
3796                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3797                         }
3798
3799                         if( $toclevel ) {
3800                                 $prevlevel = $level;
3801                                 $prevtoclevel = $toclevel;
3802                         }
3803                         $level = $matches[1][$headlineCount];
3804
3805                         if( $doNumberHeadings || $enoughToc ) {
3806
3807                                 if ( $level > $prevlevel ) {
3808                                         # Increase TOC level
3809                                         $toclevel++;
3810                                         $sublevelCount[$toclevel] = 0;
3811                                         if( $toclevel<$wgMaxTocLevel ) {
3812                                                 $prevtoclevel = $toclevel;
3813                                                 $toc .= $sk->tocIndent();
3814                                                 $numVisible++;
3815                                         }
3816                                 }
3817                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3818                                         # Decrease TOC level, find level to jump to
3819
3820                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3821                                                 # Can only go down to level 1
3822                                                 $toclevel = 1;
3823                                         } else {
3824                                                 for ($i = $toclevel; $i > 0; $i--) {
3825                                                         if ( $levelCount[$i] == $level ) {
3826                                                                 # Found last matching level
3827                                                                 $toclevel = $i;
3828                                                                 break;
3829                                                         }
3830                                                         elseif ( $levelCount[$i] < $level ) {
3831                                                                 # Found first matching level below current level
3832                                                                 $toclevel = $i + 1;
3833                                                                 break;
3834                                                         }
3835                                                 }
3836                                         }
3837                                         if( $toclevel<$wgMaxTocLevel ) {
3838                                                 if($prevtoclevel < $wgMaxTocLevel) {
3839                                                         # Unindent only if the previous toc level was shown :p
3840                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3841                                                 } else {
3842                                                         $toc .= $sk->tocLineEnd();
3843                                                 }
3844                                         }
3845                                 }
3846                                 else {
3847                                         # No change in level, end TOC line
3848                                         if( $toclevel<$wgMaxTocLevel ) {
3849                                                 $toc .= $sk->tocLineEnd();
3850                                         }
3851                                 }
3852
3853                                 $levelCount[$toclevel] = $level;
3854
3855                                 # count number of headlines for each level
3856                                 @$sublevelCount[$toclevel]++;
3857                                 $dot = 0;
3858                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3859                                         if( !empty( $sublevelCount[$i] ) ) {
3860                                                 if( $dot ) {
3861                                                         $numbering .= '.';
3862                                                 }
3863                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3864                                                 $dot = 1;
3865                                         }
3866                                 }
3867                         }
3868
3869                         # The safe header is a version of the header text safe to use for links
3870                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3871                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3872
3873                         # Remove link placeholders by the link text.
3874                         #     <!--LINK number-->
3875                         # turns into
3876                         #     link text with suffix
3877                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3878                                                             "\$this->mLinkHolders['texts'][\$1]",
3879                                                             $safeHeadline );
3880                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3881                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3882                                                             $safeHeadline );
3883
3884                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3885                         $tocline = preg_replace(
3886                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3887                                 array( '',                          '<$1>'),
3888                                 $safeHeadline
3889                         );
3890                         $tocline = trim( $tocline );
3891
3892                         # For the anchor, strip out HTML-y stuff period
3893                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3894                         $safeHeadline = trim( $safeHeadline );
3895
3896                         # Save headline for section edit hint before it's escaped
3897                         $headlineHint = $safeHeadline;
3898                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3899                         $refers[$headlineCount] = $safeHeadline;
3900
3901                         # count how many in assoc. array so we can track dupes in anchors
3902                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3903                         $refcount[$headlineCount] = $refers[$safeHeadline];
3904
3905                         # Don't number the heading if it is the only one (looks silly)
3906                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3907                                 # the two are different if the line contains a link
3908                                 $headline=$numbering . ' ' . $headline;
3909                         }
3910
3911                         # Create the anchor for linking from the TOC to the section
3912                         $anchor = $safeHeadline;
3913                         if($refcount[$headlineCount] > 1 ) {
3914                                 $anchor .= '_' . $refcount[$headlineCount];
3915                         }
3916                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3917                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3918                         }
3919                         # give headline the correct <h#> tag
3920                         if( $showEditLink && $sectionIndex !== false ) {
3921                                 if( $isTemplate )
3922                                         $editlink = $sk->editSectionLinkForOther($titleText, $sectionIndex);
3923                                 else
3924                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3925                         } else {
3926                                 $editlink = '';
3927                         }
3928                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3929
3930                         $headlineCount++;
3931                 }
3932
3933                 # Never ever show TOC if no headers
3934                 if( $numVisible < 1 ) {
3935                         $enoughToc = false;
3936                 }
3937
3938                 if( $enoughToc ) {
3939                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3940                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3941                         }
3942                         $toc = $sk->tocList( $toc );
3943                 }
3944
3945                 # split up and insert constructed headlines
3946
3947                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3948                 $i = 0;
3949
3950                 foreach( $blocks as $block ) {
3951                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3952                                 # This is the [edit] link that appears for the top block of text when
3953                                 # section editing is enabled
3954
3955                                 # Disabled because it broke block formatting
3956                                 # For example, a bullet point in the top line
3957                                 # $full .= $sk->editSectionLink(0);
3958                         }
3959                         $full .= $block;
3960                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3961                                 # Top anchor now in skin
3962                                 $full = $full.$toc;
3963                         }
3964
3965                         if( !empty( $head[$i] ) ) {
3966                                 $full .= $head[$i];
3967                         }
3968                         $i++;
3969                 }
3970                 if( $this->mForceTocPosition ) {
3971                         return str_replace( '<!--MWTOC-->', $toc, $full );
3972                 } else {
3973                         return $full;
3974                 }
3975         }
3976
3977         /**
3978          * Transform wiki markup when saving a page by doing \r\n -> \n
3979          * conversion, substitting signatures, {{subst:}} templates, etc.
3980          *
3981          * @param string $text the text to transform
3982          * @param Title &$title the Title object for the current article
3983          * @param User &$user the User object describing the current user
3984          * @param ParserOptions $options parsing options
3985          * @param bool $clearState whether to clear the parser state first
3986          * @return string the altered wiki markup
3987          * @public
3988          */
3989         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3990                 $this->mOptions = $options;
3991                 $this->mTitle =& $title;
3992                 $this->setOutputType( OT_WIKI );
3993
3994                 if ( $clearState ) {
3995                         $this->clearState();
3996                 }
3997
3998                 $pairs = array(
3999                         "\r\n" => "\n",
4000                 );
4001                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4002                 $text = $this->pstPass2( $text, $user );
4003                 $text = $this->mStripState->unstripBoth( $text );
4004                 return $text;
4005         }
4006
4007         /**
4008          * Pre-save transform helper function
4009          * @private
4010          */
4011         function pstPass2( $text, $user ) {
4012                 global $wgContLang, $wgLocaltimezone;
4013
4014                 /* Note: This is the timestamp saved as hardcoded wikitext to
4015                  * the database, we use $wgContLang here in order to give
4016                  * everyone the same signature and use the default one rather
4017                  * than the one selected in each user's preferences.
4018                  */
4019                 if ( isset( $wgLocaltimezone ) ) {
4020                         $oldtz = getenv( 'TZ' );
4021                         putenv( 'TZ='.$wgLocaltimezone );
4022                 }
4023                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4024                   ' (' . date( 'T' ) . ')';
4025                 if ( isset( $wgLocaltimezone ) ) {
4026                         putenv( 'TZ='.$oldtz );
4027                 }
4028
4029                 # Variable replacement
4030                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4031                 $text = $this->replaceVariables( $text );
4032
4033                 # Strip out <nowiki> etc. added via replaceVariables
4034                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4035
4036                 # Signatures
4037                 $sigText = $this->getUserSig( $user );
4038                 $text = strtr( $text, array(
4039                         '~~~~~' => $d,
4040                         '~~~~' => "$sigText $d",
4041                         '~~~' => $sigText
4042                 ) );
4043
4044                 # Context links: [[|name]] and [[name (context)|]]
4045                 #
4046                 global $wgLegalTitleChars;
4047                 $tc = "[$wgLegalTitleChars]";
4048                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4049
4050                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4051                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4052                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4053
4054                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4055                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4056                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4057
4058                 $t = $this->mTitle->getText();
4059                 $m = array();
4060                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4061                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4062                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4063                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4064                 } else {
4065                         # if there's no context, don't bother duplicating the title
4066                         $text = preg_replace( $p2, '[[\\1]]', $text );
4067                 }
4068
4069                 # Trim trailing whitespace
4070                 $text = rtrim( $text );
4071
4072                 return $text;
4073         }
4074
4075         /**
4076          * Fetch the user's signature text, if any, and normalize to
4077          * validated, ready-to-insert wikitext.
4078          *
4079          * @param User $user
4080          * @return string
4081          * @private
4082          */
4083         function getUserSig( &$user ) {
4084                 global $wgMaxSigChars;
4085
4086                 $username = $user->getName();
4087                 $nickname = $user->getOption( 'nickname' );
4088                 $nickname = $nickname === '' ? $username : $nickname;
4089
4090                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4091                         $nickname = $username;
4092                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4093                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4094                         # Sig. might contain markup; validate this
4095                         if( $this->validateSig( $nickname ) !== false ) {
4096                                 # Validated; clean up (if needed) and return it
4097                                 return $this->cleanSig( $nickname, true );
4098                         } else {
4099                                 # Failed to validate; fall back to the default
4100                                 $nickname = $username;
4101                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4102                         }
4103                 }
4104
4105                 // Make sure nickname doesnt get a sig in a sig
4106                 $nickname = $this->cleanSigInSig( $nickname );
4107
4108                 # If we're still here, make it a link to the user page
4109                 $userText = wfEscapeWikiText( $username );
4110                 $nickText = wfEscapeWikiText( $nickname );
4111                 if ( $user->isAnon() )  {
4112                         return wfMsgForContent( 'signature-anon', $userText, $nickText );
4113                 } else {
4114                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4115                 }
4116         }
4117
4118         /**
4119          * Check that the user's signature contains no bad XML
4120          *
4121          * @param string $text
4122          * @return mixed An expanded string, or false if invalid.
4123          */
4124         function validateSig( $text ) {
4125                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4126         }
4127
4128         /**
4129          * Clean up signature text
4130          *
4131          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4132          * 2) Substitute all transclusions
4133          *
4134          * @param string $text
4135          * @param $parsing Whether we're cleaning (preferences save) or parsing
4136          * @return string Signature text
4137          */
4138         function cleanSig( $text, $parsing = false ) {
4139                 global $wgTitle;
4140                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
4141
4142                 $substWord = MagicWord::get( 'subst' );
4143                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4144                 $substText = '{{' . $substWord->getSynonym( 0 );
4145
4146                 $text = preg_replace( $substRegex, $substText, $text );
4147                 $text = $this->cleanSigInSig( $text );
4148                 $text = $this->replaceVariables( $text );
4149
4150                 $this->clearState();
4151                 return $text;
4152         }
4153
4154         /**
4155          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4156          * @param string $text
4157          * @return string Signature text with /~{3,5}/ removed
4158          */
4159         function cleanSigInSig( $text ) {
4160                 $text = preg_replace( '/~{3,5}/', '', $text );
4161                 return $text;
4162         }
4163
4164         /**
4165          * Set up some variables which are usually set up in parse()
4166          * so that an external function can call some class members with confidence
4167          * @public
4168          */
4169         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4170                 $this->mTitle =& $title;
4171                 $this->mOptions = $options;
4172                 $this->setOutputType( $outputType );
4173                 if ( $clearState ) {
4174                         $this->clearState();
4175                 }
4176         }
4177
4178         /**
4179          * Transform a MediaWiki message by replacing magic variables.
4180          *
4181          * For some unknown reason, it also expands templates, but only to the
4182          * first recursion level. This is wrong and broken, probably introduced
4183          * accidentally during refactoring, but probably relied upon by thousands
4184          * of users.
4185          *
4186          * @param string $text the text to transform
4187          * @param ParserOptions $options  options
4188          * @return string the text with variables substituted
4189          * @public
4190          */
4191         function transformMsg( $text, $options ) {
4192                 global $wgTitle;
4193                 static $executing = false;
4194
4195                 $fname = "Parser::transformMsg";
4196
4197                 # Guard against infinite recursion
4198                 if ( $executing ) {
4199                         return $text;
4200                 }
4201                 $executing = true;
4202
4203                 wfProfileIn($fname);
4204
4205                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4206                         $this->mTitle = $wgTitle;
4207                 } else {
4208                         $this->mTitle = Title::newFromText('msg');
4209                 }
4210                 $this->mOptions = $options;
4211                 $this->setOutputType( OT_MSG );
4212                 $this->clearState();
4213                 $text = $this->replaceVariables( $text );
4214
4215                 $executing = false;
4216                 wfProfileOut($fname);
4217                 return $text;
4218         }
4219
4220         /**
4221          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4222          * The callback should have the following form:
4223          *    function myParserHook( $text, $params, &$parser ) { ... }
4224          *
4225          * Transform and return $text. Use $parser for any required context, e.g. use
4226          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4227          *
4228          * @public
4229          *
4230          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4231          * @param mixed $callback The callback function (and object) to use for the tag
4232          *
4233          * @return The old value of the mTagHooks array associated with the hook
4234          */
4235         function setHook( $tag, $callback ) {
4236                 $tag = strtolower( $tag );
4237                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4238                 $this->mTagHooks[$tag] = $callback;
4239                 $this->mStripList[] = $tag;
4240
4241                 return $oldVal;
4242         }
4243
4244         function setTransparentTagHook( $tag, $callback ) {
4245                 $tag = strtolower( $tag );
4246                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4247                 $this->mTransparentTagHooks[$tag] = $callback;
4248
4249                 return $oldVal;
4250         }
4251
4252         /**
4253          * Create a function, e.g. {{sum:1|2|3}}
4254          * The callback function should have the form:
4255          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4256          *
4257          * The callback may either return the text result of the function, or an array with the text
4258          * in element 0, and a number of flags in the other elements. The names of the flags are
4259          * specified in the keys. Valid flags are:
4260          *   found                     The text returned is valid, stop processing the template. This
4261          *                             is on by default.
4262          *   nowiki                    Wiki markup in the return value should be escaped
4263          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4264          *   noargs                    Don't replace triple-brace arguments in the return value
4265          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4266          *
4267          * @public
4268          *
4269          * @param string $id The magic word ID
4270          * @param mixed $callback The callback function (and object) to use
4271          * @param integer $flags a combination of the following flags:
4272          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4273          *
4274          * @return The old callback function for this name, if any
4275          */
4276         function setFunctionHook( $id, $callback, $flags = 0 ) {
4277                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4278                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4279
4280                 # Add to function cache
4281                 $mw = MagicWord::get( $id );
4282                 if( !$mw )
4283                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4284
4285                 $synonyms = $mw->getSynonyms();
4286                 $sensitive = intval( $mw->isCaseSensitive() );
4287
4288                 foreach ( $synonyms as $syn ) {
4289                         # Case
4290                         if ( !$sensitive ) {
4291                                 $syn = strtolower( $syn );
4292                         }
4293                         # Add leading hash
4294                         if ( !( $flags & SFH_NO_HASH ) ) {
4295                                 $syn = '#' . $syn;
4296                         }
4297                         # Remove trailing colon
4298                         if ( substr( $syn, -1, 1 ) == ':' ) {
4299                                 $syn = substr( $syn, 0, -1 );
4300                         }
4301                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4302                 }
4303                 return $oldVal;
4304         }
4305
4306         /**
4307          * Get all registered function hook identifiers
4308          *
4309          * @return array
4310          */
4311         function getFunctionHooks() {
4312                 return array_keys( $this->mFunctionHooks );
4313         }
4314
4315         /**
4316          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4317          * Placeholders created in Skin::makeLinkObj()
4318          * Returns an array of link CSS classes, indexed by PDBK.
4319          * $options is a bit field, RLH_FOR_UPDATE to select for update
4320          */
4321         function replaceLinkHolders( &$text, $options = 0 ) {
4322                 global $wgUser;
4323                 global $wgContLang;
4324
4325                 $fname = 'Parser::replaceLinkHolders';
4326                 wfProfileIn( $fname );
4327
4328                 $pdbks = array();
4329                 $colours = array();
4330                 $linkcolour_ids = array();
4331                 $sk = $this->mOptions->getSkin();
4332                 $linkCache =& LinkCache::singleton();
4333
4334                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4335                         wfProfileIn( $fname.'-check' );
4336                         $dbr = wfGetDB( DB_SLAVE );
4337                         $page = $dbr->tableName( 'page' );
4338                         $threshold = $wgUser->getOption('stubthreshold');
4339
4340                         # Sort by namespace
4341                         asort( $this->mLinkHolders['namespaces'] );
4342
4343                         # Generate query
4344                         $query = false;
4345                         $current = null;
4346                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4347                                 # Make title object
4348                                 $title = $this->mLinkHolders['titles'][$key];
4349
4350                                 # Skip invalid entries.
4351                                 # Result will be ugly, but prevents crash.
4352                                 if ( is_null( $title ) ) {
4353                                         continue;
4354                                 }
4355                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4356
4357                                 # Check if it's a static known link, e.g. interwiki
4358                                 if ( $title->isAlwaysKnown() ) {
4359                                         $colours[$pdbk] = '';
4360                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4361                                         $colours[$pdbk] = '';
4362                                         $this->mOutput->addLink( $title, $id );
4363                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4364                                         $colours[$pdbk] = 'new';
4365                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4366                                         $colours[$pdbk] = 'new';
4367                                 } else {
4368                                         # Not in the link cache, add it to the query
4369                                         if ( !isset( $current ) ) {
4370                                                 $current = $ns;
4371                                                 $query =  "SELECT page_id, page_namespace, page_title";
4372                                                 if ( $threshold > 0 ) {
4373                                                         $query .= ', page_len, page_is_redirect';
4374                                                 }
4375                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4376                                         } elseif ( $current != $ns ) {
4377                                                 $current = $ns;
4378                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4379                                         } else {
4380                                                 $query .= ', ';
4381                                         }
4382
4383                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4384                                 }
4385                         }
4386                         if ( $query ) {
4387                                 $query .= '))';
4388                                 if ( $options & RLH_FOR_UPDATE ) {
4389                                         $query .= ' FOR UPDATE';
4390                                 }
4391
4392                                 $res = $dbr->query( $query, $fname );
4393
4394                                 # Fetch data and form into an associative array
4395                                 # non-existent = broken
4396                                 while ( $s = $dbr->fetchObject($res) ) {
4397                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4398                                         $pdbk = $title->getPrefixedDBkey();
4399                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4400                                         $this->mOutput->addLink( $title, $s->page_id );
4401                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4402                                         //add id to the extension todolist
4403                                         $linkcolour_ids[$s->page_id] = $pdbk;
4404                                 }
4405                                 //pass an array of page_ids to an extension
4406                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4407                         }
4408                         wfProfileOut( $fname.'-check' );
4409
4410                         # Do a second query for different language variants of links and categories
4411                         if($wgContLang->hasVariants()){
4412                                 $linkBatch = new LinkBatch();
4413                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4414                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4415                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4416
4417                                 $categories = $this->mOutput->getCategoryLinks();
4418
4419                                 // Add variants of links to link batch
4420                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4421                                         $title = $this->mLinkHolders['titles'][$key];
4422                                         if ( is_null( $title ) )
4423                                                 continue;
4424
4425                                         $pdbk = $title->getPrefixedDBkey();
4426                                         $titleText = $title->getText();
4427
4428                                         // generate all variants of the link title text
4429                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4430
4431                                         // if link was not found (in first query), add all variants to query
4432                                         if ( !isset($colours[$pdbk]) ){
4433                                                 foreach($allTextVariants as $textVariant){
4434                                                         if($textVariant != $titleText){
4435                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4436                                                                 if(is_null($variantTitle)) continue;
4437                                                                 $linkBatch->addObj( $variantTitle );
4438                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4439                                                         }
4440                                                 }
4441                                         }
4442                                 }
4443
4444                                 // process categories, check if a category exists in some variant
4445                                 foreach( $categories as $category ){
4446                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4447                                         foreach($variants as $variant){
4448                                                 if($variant != $category){
4449                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4450                                                         if(is_null($variantTitle)) continue;
4451                                                         $linkBatch->addObj( $variantTitle );
4452                                                         $categoryMap[$variant] = $category;
4453                                                 }
4454                                         }
4455                                 }
4456
4457
4458                                 if(!$linkBatch->isEmpty()){
4459                                         // construct query
4460                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4461
4462                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4463                                         if ( $threshold > 0 ) {
4464                                                 $variantQuery .= ', page_len, page_is_redirect';
4465                                         }
4466
4467                                         $variantQuery .= " FROM $page WHERE $titleClause";
4468                                         if ( $options & RLH_FOR_UPDATE ) {
4469                                                 $variantQuery .= ' FOR UPDATE';
4470                                         }
4471
4472                                         $varRes = $dbr->query( $variantQuery, $fname );
4473
4474                                         // for each found variants, figure out link holders and replace
4475                                         while ( $s = $dbr->fetchObject($varRes) ) {
4476
4477                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4478                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4479                                                 $vardbk = $variantTitle->getDBkey();
4480
4481                                                 $holderKeys = array();
4482                                                 if(isset($variantMap[$varPdbk])){
4483                                                         $holderKeys = $variantMap[$varPdbk];
4484                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4485                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4486                                                 }
4487
4488                                                 // loop over link holders
4489                                                 foreach($holderKeys as $key){
4490                                                         $title = $this->mLinkHolders['titles'][$key];
4491                                                         if ( is_null( $title ) ) continue;
4492
4493                                                         $pdbk = $title->getPrefixedDBkey();
4494
4495                                                         if(!isset($colours[$pdbk])){
4496                                                                 // found link in some of the variants, replace the link holder data
4497                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4498                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4499
4500                                                                 // set pdbk and colour
4501                                                                 $pdbks[$key] = $varPdbk;
4502                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4503                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4504                                                         }
4505                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4506                                                 }
4507
4508                                                 // check if the object is a variant of a category
4509                                                 if(isset($categoryMap[$vardbk])){
4510                                                         $oldkey = $categoryMap[$vardbk];
4511                                                         if($oldkey != $vardbk)
4512                                                                 $varCategories[$oldkey]=$vardbk;
4513                                                 }
4514                                         }
4515
4516                                         // rebuild the categories in original order (if there are replacements)
4517                                         if(count($varCategories)>0){
4518                                                 $newCats = array();
4519                                                 $originalCats = $this->mOutput->getCategories();
4520                                                 foreach($originalCats as $cat => $sortkey){
4521                                                         // make the replacement
4522                                                         if( array_key_exists($cat,$varCategories) )
4523                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4524                                                         else $newCats[$cat] = $sortkey;
4525                                                 }
4526                                                 $this->mOutput->setCategoryLinks($newCats);
4527                                         }
4528                                 }
4529                         }
4530
4531                         # Construct search and replace arrays
4532                         wfProfileIn( $fname.'-construct' );
4533                         $replacePairs = array();
4534                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4535                                 $pdbk = $pdbks[$key];
4536                                 $searchkey = "<!--LINK $key-->";
4537                                 $title = $this->mLinkHolders['titles'][$key];
4538                                 if ( !isset( $colours[$pdbk] ) ) {
4539                                         $linkCache->addBadLinkObj( $title );
4540                                         $colours[$pdbk] = 'new';
4541                                         $this->mOutput->addLink( $title, 0 );
4542                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4543                                                                         $this->mLinkHolders['texts'][$key],
4544                                                                         $this->mLinkHolders['queries'][$key] );
4545                                 } else {
4546                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4547                                                                         $this->mLinkHolders['texts'][$key],
4548                                                                         $this->mLinkHolders['queries'][$key] );
4549                                 }
4550                         }
4551                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4552                         wfProfileOut( $fname.'-construct' );
4553
4554                         # Do the thing
4555                         wfProfileIn( $fname.'-replace' );
4556                         $text = preg_replace_callback(
4557                                 '/(<!--LINK .*?-->)/',
4558                                 $replacer->cb(),
4559                                 $text);
4560
4561                         wfProfileOut( $fname.'-replace' );
4562                 }
4563
4564                 # Now process interwiki link holders
4565                 # This is quite a bit simpler than internal links
4566                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4567                         wfProfileIn( $fname.'-interwiki' );
4568                         # Make interwiki link HTML
4569                         $replacePairs = array();
4570                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4571                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4572                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4573                         }
4574                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4575
4576                         $text = preg_replace_callback(
4577                                 '/<!--IWLINK (.*?)-->/',
4578                                 $replacer->cb(),
4579                                 $text );
4580                         wfProfileOut( $fname.'-interwiki' );
4581                 }
4582
4583                 wfProfileOut( $fname );
4584                 return $colours;
4585         }
4586
4587         /**
4588          * Replace <!--LINK--> link placeholders with plain text of links
4589          * (not HTML-formatted).
4590          * @param string $text
4591          * @return string
4592          */
4593         function replaceLinkHoldersText( $text ) {
4594                 $fname = 'Parser::replaceLinkHoldersText';
4595                 wfProfileIn( $fname );
4596
4597                 $text = preg_replace_callback(
4598                         '/<!--(LINK|IWLINK) (.*?)-->/',
4599                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4600                         $text );
4601
4602                 wfProfileOut( $fname );
4603                 return $text;
4604         }
4605
4606         /**
4607          * @param array $matches
4608          * @return string
4609          * @private
4610          */
4611         function replaceLinkHoldersTextCallback( $matches ) {
4612                 $type = $matches[1];
4613                 $key  = $matches[2];
4614                 if( $type == 'LINK' ) {
4615                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4616                                 return $this->mLinkHolders['texts'][$key];
4617                         }
4618                 } elseif( $type == 'IWLINK' ) {
4619                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4620                                 return $this->mInterwikiLinkHolders['texts'][$key];
4621                         }
4622                 }
4623                 return $matches[0];
4624         }
4625
4626         /**
4627          * Tag hook handler for 'pre'.
4628          */
4629         function renderPreTag( $text, $attribs ) {
4630                 // Backwards-compatibility hack
4631                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4632
4633                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4634                 return wfOpenElement( 'pre', $attribs ) .
4635                         Xml::escapeTagsOnly( $content ) .
4636                         '</pre>';
4637         }
4638
4639         /**
4640          * Renders an image gallery from a text with one line per image.
4641          * text labels may be given by using |-style alternative text. E.g.
4642          *   Image:one.jpg|The number "1"
4643          *   Image:tree.jpg|A tree
4644          * given as text will return the HTML of a gallery with two images,
4645          * labeled 'The number "1"' and
4646          * 'A tree'.
4647          */
4648         function renderImageGallery( $text, $params ) {
4649                 $ig = new ImageGallery();
4650                 $ig->setContextTitle( $this->mTitle );
4651                 $ig->setShowBytes( false );
4652                 $ig->setShowFilename( false );
4653                 $ig->setParser( $this );
4654                 $ig->setHideBadImages();
4655                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4656                 $ig->useSkin( $this->mOptions->getSkin() );
4657                 $ig->mRevisionId = $this->mRevisionId;
4658
4659                 if( isset( $params['caption'] ) ) {
4660                         $caption = $params['caption'];
4661                         $caption = htmlspecialchars( $caption );
4662                         $caption = $this->replaceInternalLinks( $caption );
4663                         $ig->setCaptionHtml( $caption );
4664                 }
4665                 if( isset( $params['perrow'] ) ) {
4666                         $ig->setPerRow( $params['perrow'] );
4667                 }
4668                 if( isset( $params['widths'] ) ) {
4669                         $ig->setWidths( $params['widths'] );
4670                 }
4671                 if( isset( $params['heights'] ) ) {
4672                         $ig->setHeights( $params['heights'] );
4673                 }
4674
4675                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4676
4677                 $lines = explode( "\n", $text );
4678                 foreach ( $lines as $line ) {
4679                         # match lines like these:
4680                         # Image:someimage.jpg|This is some image
4681                         $matches = array();
4682                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4683                         # Skip empty lines
4684                         if ( count( $matches ) == 0 ) {
4685                                 continue;
4686                         }
4687                         $tp = Title::newFromText( $matches[1] );
4688                         $nt =& $tp;
4689                         if( is_null( $nt ) ) {
4690                                 # Bogus title. Ignore these so we don't bomb out later.
4691                                 continue;
4692                         }
4693                         if ( isset( $matches[3] ) ) {
4694                                 $label = $matches[3];
4695                         } else {
4696                                 $label = '';
4697                         }
4698
4699                         $pout = $this->parse( $label,
4700                                 $this->mTitle,
4701                                 $this->mOptions,
4702                                 false, // Strip whitespace...?
4703                                 false  // Don't clear state!
4704                         );
4705                         $html = $pout->getText();
4706
4707                         $ig->add( $nt, $html );
4708
4709                         # Only add real images (bug #5586)
4710                         if ( $nt->getNamespace() == NS_IMAGE ) {
4711                                 $this->mOutput->addImage( $nt->getDBkey() );
4712                         }
4713                 }
4714                 return $ig->toHTML();
4715         }
4716
4717         function getImageParams( $handler ) {
4718                 if ( $handler ) {
4719                         $handlerClass = get_class( $handler );
4720                 } else {
4721                         $handlerClass = '';
4722                 }
4723                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4724                         // Initialise static lists
4725                         static $internalParamNames = array(
4726                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4727                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4728                                         'bottom', 'text-bottom' ),
4729                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4730                                         'upright', 'border' ),
4731                         );
4732                         static $internalParamMap;
4733                         if ( !$internalParamMap ) {
4734                                 $internalParamMap = array();
4735                                 foreach ( $internalParamNames as $type => $names ) {
4736                                         foreach ( $names as $name ) {
4737                                                 $magicName = str_replace( '-', '_', "img_$name" );
4738                                                 $internalParamMap[$magicName] = array( $type, $name );
4739                                         }
4740                                 }
4741                         }
4742
4743                         // Add handler params
4744                         $paramMap = $internalParamMap;
4745                         if ( $handler ) {
4746                                 $handlerParamMap = $handler->getParamMap();
4747                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4748                                         $paramMap[$magic] = array( 'handler', $paramName );
4749                                 }
4750                         }
4751                         $this->mImageParams[$handlerClass] = $paramMap;
4752                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4753                 }
4754                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4755         }
4756
4757         /**
4758          * Parse image options text and use it to make an image
4759          */
4760         function makeImage( $title, $options ) {
4761                 # @TODO: let the MediaHandler specify its transform parameters
4762                 #
4763                 # Check if the options text is of the form "options|alt text"
4764                 # Options are:
4765                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4766                 #  * left               no resizing, just left align. label is used for alt= only
4767                 #  * right              same, but right aligned
4768                 #  * none               same, but not aligned
4769                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4770                 #  * center             center the image
4771                 #  * framed             Keep original image size, no magnify-button.
4772                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4773                 #  * upright            reduce width for upright images, rounded to full __0 px
4774                 #  * border             draw a 1px border around the image
4775                 # vertical-align values (no % or length right now):
4776                 #  * baseline
4777                 #  * sub
4778                 #  * super
4779                 #  * top
4780                 #  * text-top
4781                 #  * middle
4782                 #  * bottom
4783                 #  * text-bottom
4784
4785                 $parts = array_map( 'trim', explode( '|', $options) );
4786                 $sk = $this->mOptions->getSkin();
4787
4788                 # Give extensions a chance to select the file revision for us
4789                 $skip = $time = false;
4790                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4791
4792                 if ( $skip ) {
4793                         return $sk->makeLinkObj( $title );
4794                 }
4795
4796                 # Get parameter map
4797                 $file = wfFindFile( $title, $time );
4798                 $handler = $file ? $file->getHandler() : false;
4799
4800                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4801
4802                 # Process the input parameters
4803                 $caption = '';
4804                 $params = array( 'frame' => array(), 'handler' => array(),
4805                         'horizAlign' => array(), 'vertAlign' => array() );
4806                 foreach( $parts as $part ) {
4807                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4808                         if ( isset( $paramMap[$magicName] ) ) {
4809                                 list( $type, $paramName ) = $paramMap[$magicName];
4810                                 $params[$type][$paramName] = $value;
4811
4812                                 // Special case; width and height come in one variable together
4813                                 if( $type == 'handler' && $paramName == 'width' ) {
4814                                         $m = array();
4815                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4816                                                 $params[$type]['width'] = intval( $m[1] );
4817                                                 $params[$type]['height'] = intval( $m[2] );
4818                                         } else {
4819                                                 $params[$type]['width'] = intval( $value );
4820                                         }
4821                                 }
4822                         } else {
4823                                 $caption = $part;
4824                         }
4825                 }
4826
4827                 # Process alignment parameters
4828                 if ( $params['horizAlign'] ) {
4829                         $params['frame']['align'] = key( $params['horizAlign'] );
4830                 }
4831                 if ( $params['vertAlign'] ) {
4832                         $params['frame']['valign'] = key( $params['vertAlign'] );
4833                 }
4834
4835                 # Validate the handler parameters
4836                 if ( $handler ) {
4837                         foreach ( $params['handler'] as $name => $value ) {
4838                                 if ( !$handler->validateParam( $name, $value ) ) {
4839                                         unset( $params['handler'][$name] );
4840                                 }
4841                         }
4842                 }
4843
4844                 # Strip bad stuff out of the alt text
4845                 $alt = $this->replaceLinkHoldersText( $caption );
4846
4847                 # make sure there are no placeholders in thumbnail attributes
4848                 # that are later expanded to html- so expand them now and
4849                 # remove the tags
4850                 $alt = $this->mStripState->unstripBoth( $alt );
4851                 $alt = Sanitizer::stripAllTags( $alt );
4852
4853                 $params['frame']['alt'] = $alt;
4854                 $params['frame']['caption'] = $caption;
4855
4856                 # Linker does the rest
4857                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4858
4859                 # Give the handler a chance to modify the parser object
4860                 if ( $handler ) {
4861                         $handler->parserTransformHook( $this, $file );
4862                 }
4863
4864                 return $ret;
4865         }
4866
4867         /**
4868          * Set a flag in the output object indicating that the content is dynamic and
4869          * shouldn't be cached.
4870          */
4871         function disableCache() {
4872                 wfDebug( "Parser output marked as uncacheable.\n" );
4873                 $this->mOutput->mCacheTime = -1;
4874         }
4875
4876         /**#@+
4877          * Callback from the Sanitizer for expanding items found in HTML attribute
4878          * values, so they can be safely tested and escaped.
4879          * @param string $text
4880          * @param PPFrame $frame
4881          * @return string
4882          * @private
4883          */
4884         function attributeStripCallback( &$text, $frame = false ) {
4885                 $text = $this->replaceVariables( $text, $frame );
4886                 $text = $this->mStripState->unstripBoth( $text );
4887                 return $text;
4888         }
4889
4890         /**#@-*/
4891
4892         /**#@+
4893          * Accessor/mutator
4894          */
4895         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4896         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4897         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4898         /**#@-*/
4899
4900         /**#@+
4901          * Accessor
4902          */
4903         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4904         /**#@-*/
4905
4906
4907         /**
4908          * Break wikitext input into sections, and either pull or replace
4909          * some particular section's text.
4910          *
4911          * External callers should use the getSection and replaceSection methods.
4912          *
4913          * @param $text Page wikitext
4914          * @param $section Numbered section. 0 pulls the text before the first
4915          *                 heading; other numbers will pull the given section
4916          *                 along with its lower-level subsections. If the section is
4917          *                 not found, $mode=get will return $newtext, and
4918          *                 $mode=replace will return $text.
4919          * @param $mode One of "get" or "replace"
4920          * @param $newText Replacement text for section data.
4921          * @return string for "get", the extracted section text.
4922          *                for "replace", the whole page with the section replaced.
4923          */
4924         private function extractSections( $text, $section, $mode, $newText='' ) {
4925                 $this->clearState();
4926                 $this->mOptions = new ParserOptions;
4927                 $this->setOutputType( OT_WIKI );
4928                 $curIndex = 0;
4929                 $outText = '';
4930                 $frame = new PPFrame( $this );
4931
4932                 // Preprocess the text
4933                 $dom = $this->preprocessToDom( $text );
4934                 $root = $dom->documentElement;
4935
4936                 // <h> nodes indicate section breaks
4937                 // They can only occur at the top level, so we can find them by iterating the root's children
4938                 $node = $root->firstChild;
4939
4940                 // Find the target section
4941                 if ( $section == 0 ) {
4942                         // Section zero doesn't nest, level=big
4943                         $targetLevel = 1000;
4944                 } else {
4945                         while ( $node ) {
4946                                 if ( $node->nodeName == 'h' ) {
4947                                         if ( $curIndex + 1 == $section ) {
4948                                                 break;
4949                                         }
4950                                         $curIndex++;
4951                                 }
4952                                 if ( $mode == 'replace' ) {
4953                                         $outText .= $frame->expand( $node );
4954                                 }
4955                                 $node = $node->nextSibling;
4956                         }
4957                         if ( $node ) {
4958                                 $targetLevel = $node->getAttribute( 'level' );
4959                         }
4960                 }
4961
4962                 if ( !$node ) {
4963                         // Not found
4964                         if ( $mode == 'get' ) {
4965                                 return $newText;
4966                         } else {
4967                                 return $text;
4968                         }
4969                 }
4970
4971                 // Find the end of the section, including nested sections
4972                 do {
4973                         if ( $node->nodeName == 'h' ) {
4974                                 $curIndex++;
4975                                 $curLevel = $node->getAttribute( 'level' );
4976                                 if ( $curIndex != $section && $curLevel <= $targetLevel ) {
4977                                         break;
4978                                 }
4979                         }
4980                         if ( $mode == 'get' ) {
4981                                 $outText .= $frame->expand( $node );
4982                         }
4983                         $node = $node->nextSibling;
4984                 } while ( $node );
4985
4986                 // Write out the remainder (in replace mode only)
4987                 if ( $mode == 'replace' ) {
4988                         // Output the replacement text
4989                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4990                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4991                         $outText .= $newText . "\n\n";
4992                         while ( $node ) {
4993                                 $outText .= $frame->expand( $node );
4994                                 $node = $node->nextSibling;
4995                         }
4996                 }
4997
4998                 if ( is_string( $outText ) ) {
4999                         // Re-insert stripped tags
5000                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5001                 }
5002
5003                 return $outText;
5004         }
5005
5006         /**
5007          * This function returns the text of a section, specified by a number ($section).
5008          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5009          * the first section before any such heading (section 0).
5010          *
5011          * If a section contains subsections, these are also returned.
5012          *
5013          * @param $text String: text to look in
5014          * @param $section Integer: section number
5015          * @param $deftext: default to return if section is not found
5016          * @return string text of the requested section
5017          */
5018         public function getSection( $text, $section, $deftext='' ) {
5019                 return $this->extractSections( $text, $section, "get", $deftext );
5020         }
5021
5022         public function replaceSection( $oldtext, $section, $text ) {
5023                 return $this->extractSections( $oldtext, $section, "replace", $text );
5024         }
5025
5026         /**
5027          * Get the timestamp associated with the current revision, adjusted for
5028          * the default server-local timestamp
5029          */
5030         function getRevisionTimestamp() {
5031                 if ( is_null( $this->mRevisionTimestamp ) ) {
5032                         wfProfileIn( __METHOD__ );
5033                         global $wgContLang;
5034                         $dbr = wfGetDB( DB_SLAVE );
5035                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5036                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5037
5038                         // Normalize timestamp to internal MW format for timezone processing.
5039                         // This has the added side-effect of replacing a null value with
5040                         // the current time, which gives us more sensible behavior for
5041                         // previews.
5042                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5043
5044                         // The cryptic '' timezone parameter tells to use the site-default
5045                         // timezone offset instead of the user settings.
5046                         //
5047                         // Since this value will be saved into the parser cache, served
5048                         // to other users, and potentially even used inside links and such,
5049                         // it needs to be consistent for all visitors.
5050                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5051
5052                         wfProfileOut( __METHOD__ );
5053                 }
5054                 return $this->mRevisionTimestamp;
5055         }
5056
5057         /**
5058          * Mutator for $mDefaultSort
5059          *
5060          * @param $sort New value
5061          */
5062         public function setDefaultSort( $sort ) {
5063                 $this->mDefaultSort = $sort;
5064         }
5065
5066         /**
5067          * Accessor for $mDefaultSort
5068          * Will use the title/prefixed title if none is set
5069          *
5070          * @return string
5071          */
5072         public function getDefaultSort() {
5073                 if( $this->mDefaultSort !== false ) {
5074                         return $this->mDefaultSort;
5075                 } else {
5076                         return $this->mTitle->getNamespace() == NS_CATEGORY
5077                                         ? $this->mTitle->getText()
5078                                         : $this->mTitle->getPrefixedText();
5079                 }
5080         }
5081
5082         /**
5083          * Try to guess the section anchor name based on a wikitext fragment
5084          * presumably extracted from a heading, for example "Header" from
5085          * "== Header ==".
5086          */
5087         public function guessSectionNameFromWikiText( $text ) {
5088                 # Strip out wikitext links(they break the anchor)
5089                 $text = $this->stripSectionName( $text );
5090                 $headline = Sanitizer::decodeCharReferences( $text );
5091                 # strip out HTML
5092                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5093                 $headline = trim( $headline );
5094                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5095                 $replacearray = array(
5096                         '%3A' => ':',
5097                         '%' => '.'
5098                 );
5099                 return str_replace(
5100                         array_keys( $replacearray ),
5101                         array_values( $replacearray ),
5102                         $sectionanchor );
5103         }
5104
5105         /**
5106          * Strips a text string of wikitext for use in a section anchor
5107          *
5108          * Accepts a text string and then removes all wikitext from the
5109          * string and leaves only the resultant text (i.e. the result of
5110          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5111          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5112          * to create valid section anchors by mimicing the output of the
5113          * parser when headings are parsed.
5114          *
5115          * @param $text string Text string to be stripped of wikitext
5116          * for use in a Section anchor
5117          * @return Filtered text string
5118          */
5119         public function stripSectionName( $text ) {
5120                 # Strip internal link markup
5121                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5122                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5123
5124                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5125                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5126                 # on how many empty links there are on the page - need to figure that out.
5127                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5128
5129                 # Parse wikitext quotes (italics & bold)
5130                 $text = $this->doQuotes($text);
5131
5132                 # Strip HTML tags
5133                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5134                 return $text;
5135         }
5136
5137         /**
5138          * strip/replaceVariables/unstrip for preprocessor regression testing
5139          */
5140         function srvus( $text ) {
5141                 $text = $this->replaceVariables( $text );
5142                 $text = $this->mStripState->unstripBoth( $text );
5143                 return $text;
5144         }
5145 }
5146
5147 /**
5148  * @todo document, briefly.
5149  * @addtogroup Parser
5150  */
5151 class OnlyIncludeReplacer {
5152         var $output = '';
5153
5154         function replace( $matches ) {
5155                 if ( substr( $matches[1], -1 ) == "\n" ) {
5156                         $this->output .= substr( $matches[1], 0, -1 );
5157                 } else {
5158                         $this->output .= $matches[1];
5159                 }
5160         }
5161 }
5162
5163 /**
5164  * @todo document, briefly.
5165  * @addtogroup Parser
5166  */
5167 class StripState {
5168         var $general, $nowiki;
5169
5170         function __construct() {
5171                 $this->general = new ReplacementArray;
5172                 $this->nowiki = new ReplacementArray;
5173         }
5174
5175         function unstripGeneral( $text ) {
5176                 wfProfileIn( __METHOD__ );
5177                 do {
5178                         $oldText = $text;
5179                         $text = $this->general->replace( $text );
5180                 } while ( $text != $oldText );
5181                 wfProfileOut( __METHOD__ );
5182                 return $text;
5183         }
5184
5185         function unstripNoWiki( $text ) {
5186                 wfProfileIn( __METHOD__ );
5187                 do {
5188                         $oldText = $text;
5189                         $text = $this->nowiki->replace( $text );
5190                 } while ( $text != $oldText );
5191                 wfProfileOut( __METHOD__ );
5192                 return $text;
5193         }
5194
5195         function unstripBoth( $text ) {
5196                 wfProfileIn( __METHOD__ );
5197                 do {
5198                         $oldText = $text;
5199                         $text = $this->general->replace( $text );
5200                         $text = $this->nowiki->replace( $text );
5201                 } while ( $text != $oldText );
5202                 wfProfileOut( __METHOD__ );
5203                 return $text;
5204         }
5205 }
5206
5207 /**
5208  * An expansion frame, used as a context to expand the result of preprocessToDom()
5209  */
5210 class PPFrame {
5211         var $parser, $title;
5212
5213         const NO_ARGS = 1;
5214         const NO_TEMPLATES = 2;
5215
5216         /**
5217          * Construct a new preprocessor frame.
5218          * @param Parser $parser The parent parser
5219          * @param Title $title The context title, or false if there isn't one
5220          */
5221         function __construct( $parser ) {
5222                 $this->parser = $parser;
5223                 $this->title = $parser->mTitle;
5224         }
5225
5226         /**
5227          * Create a new child frame
5228          * $args is optionally a DOMNodeList containing the template arguments
5229          */
5230         function newChild( $args = false, $title = false ) {
5231                 $assocArgs = array();
5232                 if ( $title === false ) {
5233                         $title = $this->title;
5234                 }
5235                 if ( $args !== false ) {
5236                         $xpath = false;
5237                         foreach ( $args as $arg ) {
5238                                 if ( !$xpath ) {
5239                                         $xpath = new DOMXPath( $arg->ownerDocument );
5240                                 }
5241
5242                                 $nameNodes = $xpath->query( 'name', $arg );
5243                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5244                                         // Numbered parameter
5245                                         $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5246                                 } else {
5247                                         // Named parameter
5248                                         $name = $this->expand( $nameNodes->item( 0 ) );
5249                                 }
5250
5251                                 $value = $xpath->query( 'value', $arg );
5252                                 $assocArgs[$name] = $value->item( 0 );
5253                         }
5254                 }
5255                 return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
5256         }
5257
5258         /**
5259          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5260          * using the current context
5261          * @param $root the node
5262          */
5263         function expand( $root, $shallowFlags = 0, $deepFlags = 0 ) {
5264                 if ( is_string( $root ) ) {
5265                         return $root;
5266                 }
5267
5268                 if ( $this->parser->ot['html']
5269                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5270                 {
5271                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5272                 }
5273                 $flags = $shallowFlags | $deepFlags;
5274
5275                 if ( is_array( $root ) ) {
5276                         $s = '';
5277                         foreach ( $root as $node ) {
5278                                 $s .= $this->expand( $node, 0, $deepFlags );
5279                         }
5280                 } elseif ( $root instanceof DOMNodeList ) {
5281                         $s = '';
5282                         foreach ( $root as $node ) {
5283                                 $s .= $this->expand( $node, 0, $deepFlags );
5284                         }
5285                 } elseif ( $root instanceof DOMNode ) {
5286                         if ( $root->nodeType == XML_TEXT_NODE ) {
5287                                 $s = $root->nodeValue;
5288                         } elseif ( $root->nodeName == 'template' ) {
5289                                 # Double-brace expansion
5290                                 $xpath = new DOMXPath( $root->ownerDocument );
5291                                 $titles = $xpath->query( 'title', $root );
5292                                 $title = $titles->item( 0 );
5293                                 $parts = $xpath->query( 'part', $root );
5294                                 if ( $flags & self::NO_TEMPLATES ) {
5295                                         $s = '{{' . $this->implodeWithFlags( '|', 0, $deepFlags, $title, $parts ) . '}}';
5296                                 } else {
5297                                         $lineStart = $root->getAttribute( 'lineStart' );
5298                                         $params = array(
5299                                                 'title' => $title,
5300                                                 'parts' => $parts,
5301                                                 'lineStart' => $lineStart,
5302                                                 'text' => 'FIXME' );
5303                                         $s = $this->parser->braceSubstitution( $params, $this );
5304                                 }
5305                         } elseif ( $root->nodeName == 'tplarg' ) {
5306                                 # Triple-brace expansion
5307                                 $xpath = new DOMXPath( $root->ownerDocument );
5308                                 $titles = $xpath->query( 'title', $root );
5309                                 $title = $titles->item( 0 );
5310                                 $parts = $xpath->query( 'part', $root );
5311                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5312                                         $s = '{{{' . $this->implode( '|', 0, $deepFlags, $title, $parts ) . '}}}';
5313                                 } else {
5314                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5315                                         $s = $this->parser->argSubstitution( $params, $this );
5316                                 }
5317                         } elseif ( $root->nodeName == 'ext' ) {
5318                                 # Extension tag
5319                                 $xpath = new DOMXPath( $root->ownerDocument );
5320                                 $names = $xpath->query( 'name', $root );
5321                                 $attrs = $xpath->query( 'attr', $root );
5322                                 $inners = $xpath->query( 'inner', $root );
5323                                 $closes = $xpath->query( 'close', $root );
5324                                 $params = array(
5325                                         'name' => $names->item( 0 ),
5326                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5327                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5328                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5329                                 );
5330                                 $s = $this->parser->extensionSubstitution( $params, $this );
5331                         } elseif ( $root->nodeName == 'h' ) {
5332                                 # Heading
5333                                 $s = $this->expand( $root->childNodes, 0, $deepFlags );
5334
5335                                 if ( $this->parser->ot['html'] ) {
5336                                         # Insert heading index marker
5337                                         $headingIndex = $root->getAttribute( 'i' );
5338                                         $titleText = $this->title->getPrefixedDBkey();
5339                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5340                                         $serial = count( $this->parser->mHeadings ) - 1;
5341                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5342                                         $count = $root->getAttribute( 'level' );
5343
5344                                         // FIXME: bug-for-bug with old parser
5345                                         // Lose whitespace for no apparent reason
5346                                         // Remove this after differential testing is done
5347                                         if ( true ) {
5348                                                 // Good version
5349                                                 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5350                                         } else {
5351                                                 // Bad version
5352                                                 if ( preg_match( '/^(={1,6})(.*?)(={1,6})\s*?$/', $s, $m ) ) {
5353                                                         if ( $m[2] != '' ) {
5354                                                                 $s = $m[1] . $marker . $m[2] . $m[3];
5355                                                         }
5356                                                 }
5357                                         }
5358                                         $this->parser->mStripState->general->setPair( $marker, '' );
5359                                 }
5360                         } else {
5361                                 # Generic recursive expansion
5362                                 $s = '';
5363                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5364                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5365                                                 $s .= $node->nodeValue;
5366                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5367                                                 $s .= $this->expand( $node, 0, $deepFlags );
5368                                         }
5369                                 }
5370                         }
5371                 } else {
5372                         throw new MWException( __METHOD__.': Invalid parameter type' );
5373                 }
5374                 return $s;
5375         }
5376
5377         function implodeWithFlags( $sep, $shallowFlags, $deepFlags /*, ... */ ) {
5378                 $args = array_slice( func_get_args(), 3 );
5379
5380                 $first = true;
5381                 $s = '';
5382                 foreach ( $args as $root ) {
5383                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5384                                 $root = array( $root );
5385                         }
5386                         foreach ( $root as $node ) {
5387                                 if ( $first ) {
5388                                         $first = false;
5389                                 } else {
5390                                         $s .= $sep;
5391                                 }
5392                                 $s .= $this->expand( $node, $shallowFlags, $deepFlags );
5393                         }
5394                 }
5395                 return $s;
5396         }
5397
5398         function implode( $sep /*, ... */ ) {
5399                 $args = func_get_args();
5400                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0, 0 ), array_slice( $args, 1 ) );
5401                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5402         }
5403
5404         function __toString() {
5405                 return 'frame{}';
5406         }
5407 }
5408
5409 /**
5410  * Expansion frame with template arguments
5411  */
5412 class PPTemplateFrame extends PPFrame {
5413         public $parser, $args, $parent, $serial;
5414
5415         function __construct( $parser, $parent = false, $args = array(), $title = false ) {
5416                 $this->parser = $parser;
5417                 $this->parent = $parent;
5418                 $this->args = $args;
5419                 $this->title = $title;
5420         }
5421
5422         function __toString() {
5423                 $s = 'tplframe{';
5424                 $first = true;
5425                 foreach ( $this->args as $name => $value ) {
5426                         if ( $first ) {
5427                                 $first = false;
5428                         } else {
5429                                 $s .= ', ';
5430                         }
5431                         $s .= "\"$name\":\"" .
5432                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5433                 }
5434                 $s .= '}';
5435                 return $s;
5436         }
5437 }
5438