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  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.2';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)\\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/S';
  62
  63         // State constants for the definition list colon extraction
  64         const COLON_STATE_TEXT = 0;
  65         const COLON_STATE_TAG = 1;
  66         const COLON_STATE_TAGSTART = 2;
  67         const COLON_STATE_CLOSETAG = 3;
  68         const COLON_STATE_TAGSLASH = 4;
  69         const COLON_STATE_COMMENT = 5;
  70         const COLON_STATE_COMMENTDASH = 6;
  71         const COLON_STATE_COMMENTDASHDASH = 7;
  72
  73         /**#@+
  74          * @private
  75          */
  76         # Persistent:
  77         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  78                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  79                 $mExtLinkBracketedRegex;
  80
  81         # Cleared with clearState():
  82         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  83         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  84         var $mInterwikiLinkHolders, $mLinkHolders;
  85         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  86         var $mTplExpandCache,// empty-frame expansion cache
  87             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  88                                 // in this path. Used for loop detection.
  89         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  90
  91         # Temporary
  92         # These are variables reset at least once per parse regardless of $clearState
  93         var $mOptions,      // ParserOptions object
  94                 $mTitle,        // Title context, used for self-link rendering and similar things
  95                 $mOutputType,   // Output type, one of the OT_xxx constants
  96                 $ot,            // Shortcut alias, see setOutputType()
  97                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  98                 $mRevisionTimestamp, // The timestamp of the specified revision ID
  99                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 100
 101         /**#@-*/
 102
 103         /**
 104          * Constructor
 105          *
 106          * @public
 107          */
 108         function __construct( $conf = array() ) {
 109                 $this->mTagHooks = array();
 110                 $this->mTransparentTagHooks = array();
 111                 $this->mFunctionHooks = array();
 112                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 113                 $this->mStripList = array( 'nowiki', 'gallery' );
 114                 $this->mMarkerSuffix = "-QINU\x7f";
 115                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 116                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 117                 $this->mFirstCall = true;
 118         }
 119
 120         /**
 121          * Do various kinds of initialisation on the first call of the parser
 122          */
 123         function firstCallInit() {
 124                 if ( !$this->mFirstCall ) {
 125                         return;
 126                 }
 127
 128                 wfProfileIn( __METHOD__ );
 129                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 130
 131                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 132
 133                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 134                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 135                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 136                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 137                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 138                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 139                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 140                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 141                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 142                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 158                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'filepath', array( 'CoreParserFunctions', 'filepath' ), SFH_NO_HASH );
 160
 161                 if ( $wgAllowDisplayTitle ) {
 162                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 163                 }
 164                 if ( $wgAllowSlowParserFunctions ) {
 165                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 166                 }
 167
 168                 $this->initialiseVariables();
 169                 $this->mFirstCall = false;
 170                 wfProfileOut( __METHOD__ );
 171         }
 172
 173         /**
 174          * Clear Parser state
 175          *
 176          * @private
 177          */
 178         function clearState() {
 179                 wfProfileIn( __METHOD__ );
 180                 if ( $this->mFirstCall ) {
 181                         $this->firstCallInit();
 182                 }
 183                 $this->mOutput = new ParserOutput;
 184                 $this->mAutonumber = 0;
 185                 $this->mLastSection = '';
 186                 $this->mDTopen = false;
 187                 $this->mIncludeCount = array();
 188                 $this->mStripState = new StripState;
 189                 $this->mArgStack = false;
 190                 $this->mInPre = false;
 191                 $this->mInterwikiLinkHolders = array(
 192                         'texts' => array(),
 193                         'titles' => array()
 194                 );
 195                 $this->mLinkHolders = array(
 196                         'namespaces' => array(),
 197                         'dbkeys' => array(),
 198                         'queries' => array(),
 199                         'texts' => array(),
 200                         'titles' => array()
 201                 );
 202                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 203
 204                 /**
 205                  * Prefix for temporary replacement strings for the multipass parser.
 206                  * \x07 should never appear in input as it's disallowed in XML.
 207                  * Using it at the front also gives us a little extra robustness
 208                  * since it shouldn't match when butted up against identifier-like
 209                  * string constructs.
 210                  *
 211                  * Must not consist of all title characters, or else it will change
 212                  * the behaviour of <nowiki> in a link.
 213                  */
 214                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 215                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 216
 217                 # Clear these on every parse, bug 4549
 218                 $this->mTemplatePath = array();
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230
 231                 wfRunHooks( 'ParserClearState', array( &$this ) );
 232                 wfProfileOut( __METHOD__ );
 233         }
 234
 235         function setOutputType( $ot ) {
 236                 $this->mOutputType = $ot;
 237                 // Shortcut alias
 238                 $this->ot = array(
 239                         'html' => $ot == OT_HTML,
 240                         'wiki' => $ot == OT_WIKI,
 241                         'msg' => $ot == OT_MSG,
 242                         'pre' => $ot == OT_PREPROCESS,
 243                 );
 244         }
 245
 246         /**
 247          * Accessor for mUniqPrefix.
 248          *
 249          * @public
 250          */
 251         function uniqPrefix() {
 252                 return $this->mUniqPrefix;
 253         }
 254
 255         /**
 256          * Convert wikitext to HTML
 257          * Do not call this function recursively.
 258          *
 259          * @param string $text Text we want to parse
 260          * @param Title &$title A title object
 261          * @param array $options
 262          * @param boolean $linestart
 263          * @param boolean $clearState
 264          * @param int $revid number to pass in {{REVISIONID}}
 265          * @return ParserOutput a ParserOutput
 266          */
 267         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 268                 /**
 269                  * First pass--just handle <nowiki> sections, pass the rest off
 270                  * to internalParse() which does all the real work.
 271                  */
 272
 273                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 274                 $fname = 'Parser::parse-' . wfGetCaller();
 275                 wfProfileIn( __METHOD__ );
 276                 wfProfileIn( $fname );
 277
 278                 if ( $clearState ) {
 279                         $this->clearState();
 280                 }
 281
 282                 $this->mOptions = $options;
 283                 $this->mTitle =& $title;
 284                 $oldRevisionId = $this->mRevisionId;
 285                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 286                 if( $revid !== null ) {
 287                         $this->mRevisionId = $revid;
 288                         $this->mRevisionTimestamp = null;
 289                 }
 290                 $this->setOutputType( OT_HTML );
 291                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 292                 # No more strip!
 293                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 294                 $text = $this->internalParse( $text );
 295                 $text = $this->mStripState->unstripGeneral( $text );
 296
 297                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 298                 $fixtags = array(
 299                         # french spaces, last one Guillemet-left
 300                         # only if there is something before the space
 301                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 302                         # french spaces, Guillemet-right
 303                         '/(\\302\\253) /' => '\\1&nbsp;',
 304                 );
 305                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 306
 307                 # only once and last
 308                 $text = $this->doBlockLevels( $text, $linestart );
 309
 310                 $this->replaceLinkHolders( $text );
 311
 312                 # the position of the parserConvert() call should not be changed. it
 313                 # assumes that the links are all replaced and the only thing left
 314                 # is the <nowiki> mark.
 315                 # Side-effects: this calls $this->mOutput->setTitleText()
 316                 $text = $wgContLang->parserConvert( $text, $this );
 317
 318                 $text = $this->mStripState->unstripNoWiki( $text );
 319
 320                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 321
 322 //!JF Move to its own function
 323
 324                 $uniq_prefix = $this->mUniqPrefix;
 325                 $matches = array();
 326                 $elements = array_keys( $this->mTransparentTagHooks );
 327                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 328
 329                 foreach( $matches as $marker => $data ) {
 330                         list( $element, $content, $params, $tag ) = $data;
 331                         $tagName = strtolower( $element );
 332                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 333                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 334                                         array( $content, $params, $this ) );
 335                         } else {
 336                                 $output = $tag;
 337                         }
 338                         $this->mStripState->general->setPair( $marker, $output );
 339                 }
 340                 $text = $this->mStripState->unstripGeneral( $text );
 341
 342                 $text = Sanitizer::normalizeCharReferences( $text );
 343
 344                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 345                         $text = Parser::tidy($text);
 346                 } else {
 347                         # attempt to sanitize at least some nesting problems
 348                         # (bug #2702 and quite a few others)
 349                         $tidyregs = array(
 350                                 # ''Something [http://www.cool.com cool''] -->
 351                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 352                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 353                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 354                                 # fix up an anchor inside another anchor, only
 355                                 # at least for a single single nested link (bug 3695)
 356                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 357                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 358                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 359                                 # contains a div, so fix it up here; replace
 360                                 # div with escaped text
 361                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 362                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 363                                 # remove empty italic or bold tag pairs, some
 364                                 # introduced by rules above
 365                                 '/<([bi])><\/\\1>/' => '',
 366                         );
 367
 368                         $text = preg_replace(
 369                                 array_keys( $tidyregs ),
 370                                 array_values( $tidyregs ),
 371                                 $text );
 372                 }
 373
 374                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 375
 376                 # Information on include size limits, for the benefit of users who try to skirt them
 377                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 378                         $max = $this->mOptions->getMaxIncludeSize();
 379                         $text .= "<!-- \n" .
 380                                 "Preprocessor node count: {$this->mPPNodeCount}\n" .
 381                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 382                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 383                                 "Maximum: $max bytes\n" .
 384                                 "-->\n";
 385                 }
 386                 $this->mOutput->setText( $text );
 387                 $this->mRevisionId = $oldRevisionId;
 388                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 389                 wfProfileOut( $fname );
 390                 wfProfileOut( __METHOD__ );
 391
 392                 return $this->mOutput;
 393         }
 394
 395         /**
 396          * Recursive parser entry point that can be called from an extension tag
 397          * hook.
 398          */
 399         function recursiveTagParse( $text ) {
 400                 wfProfileIn( __METHOD__ );
 401                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 402                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 403                 $text = $this->internalParse( $text );
 404                 wfProfileOut( __METHOD__ );
 405                 return $text;
 406         }
 407
 408         /**
 409          * Expand templates and variables in the text, producing valid, static wikitext.
 410          * Also removes comments.
 411          */
 412         function preprocess( $text, $title, $options, $revid = null ) {
 413                 wfProfileIn( __METHOD__ );
 414                 $this->clearState();
 415                 $this->setOutputType( OT_PREPROCESS );
 416                 $this->mOptions = $options;
 417                 $this->mTitle = $title;
 418                 if( $revid !== null ) {
 419                         $this->mRevisionId = $revid;
 420                 }
 421                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 422                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 423                 $text = $this->replaceVariables( $text );
 424                 if ( $this->mOptions->getRemoveComments() ) {
 425                         $text = Sanitizer::removeHTMLcomments( $text );
 426                 }
 427                 $text = $this->mStripState->unstripBoth( $text );
 428                 wfProfileOut( __METHOD__ );
 429                 return $text;
 430         }
 431
 432         /**
 433          * Get a random string
 434          *
 435          * @private
 436          * @static
 437          */
 438         function getRandomString() {
 439                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 440         }
 441
 442         function &getTitle() { return $this->mTitle; }
 443         function getOptions() { return $this->mOptions; }
 444
 445         function getFunctionLang() {
 446                 global $wgLang, $wgContLang;
 447                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 448         }
 449
 450         /**
 451          * Replaces all occurrences of HTML-style comments and the given tags
 452          * in the text with a random marker and returns teh next text. The output
 453          * parameter $matches will be an associative array filled with data in
 454          * the form:
 455          *   'UNIQ-xxxxx' => array(
 456          *     'element',
 457          *     'tag content',
 458          *     array( 'param' => 'x' ),
 459          *     '<element param="x">tag content</element>' ) )
 460          *
 461          * @param $elements list of element names. Comments are always extracted.
 462          * @param $text Source text string.
 463          * @param $uniq_prefix
 464          *
 465          * @public
 466          * @static
 467          */
 468         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 469                 static $n = 1;
 470                 $stripped = '';
 471                 $matches = array();
 472
 473                 $taglist = implode( '|', $elements );
 474                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 475
 476                 while ( '' != $text ) {
 477                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 478                         $stripped .= $p[0];
 479                         if( count( $p ) < 5 ) {
 480                                 break;
 481                         }
 482                         if( count( $p ) > 5 ) {
 483                                 // comment
 484                                 $element    = $p[4];
 485                                 $attributes = '';
 486                                 $close      = '';
 487                                 $inside     = $p[5];
 488                         } else {
 489                                 // tag
 490                                 $element    = $p[1];
 491                                 $attributes = $p[2];
 492                                 $close      = $p[3];
 493                                 $inside     = $p[4];
 494                         }
 495
 496                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 497                         $stripped .= $marker;
 498
 499                         if ( $close === '/>' ) {
 500                                 // Empty element tag, <tag />
 501                                 $content = null;
 502                                 $text = $inside;
 503                                 $tail = null;
 504                         } else {
 505                                 if( $element == '!--' ) {
 506                                         $end = '/(-->)/';
 507                                 } else {
 508                                         $end = "/(<\\/$element\\s*>)/i";
 509                                 }
 510                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 511                                 $content = $q[0];
 512                                 if( count( $q ) < 3 ) {
 513                                         # No end tag -- let it run out to the end of the text.
 514                                         $tail = '';
 515                                         $text = '';
 516                                 } else {
 517                                         $tail = $q[1];
 518                                         $text = $q[2];
 519                                 }
 520                         }
 521
 522                         $matches[$marker] = array( $element,
 523                                 $content,
 524                                 Sanitizer::decodeTagAttributes( $attributes ),
 525                                 "<$element$attributes$close$content$tail" );
 526                 }
 527                 return $stripped;
 528         }
 529
 530         /**
 531          * Get a list of strippable XML-like elements
 532          */
 533         function getStripList() {
 534                 global $wgRawHtml;
 535                 $elements = $this->mStripList;
 536                 if( $wgRawHtml ) {
 537                         $elements[] = 'html';
 538                 }
 539                 if( $this->mOptions->getUseTeX() ) {
 540                         $elements[] = 'math';
 541                 }
 542                 return $elements;
 543         }
 544
 545         /**
 546          * @deprecated use replaceVariables
 547          */
 548         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 549                 return $text;
 550         }
 551
 552         /**
 553          * Restores pre, math, and other extensions removed by strip()
 554          *
 555          * always call unstripNoWiki() after this one
 556          * @private
 557          * @deprecated use $this->mStripState->unstrip()
 558          */
 559         function unstrip( $text, $state ) {
 560                 return $state->unstripGeneral( $text );
 561         }
 562
 563         /**
 564          * Always call this after unstrip() to preserve the order
 565          *
 566          * @private
 567          * @deprecated use $this->mStripState->unstrip()
 568          */
 569         function unstripNoWiki( $text, $state ) {
 570                 return $state->unstripNoWiki( $text );
 571         }
 572
 573         /**
 574          * @deprecated use $this->mStripState->unstripBoth()
 575          */
 576         function unstripForHTML( $text ) {
 577                 return $this->mStripState->unstripBoth( $text );
 578         }
 579
 580         /**
 581          * Add an item to the strip state
 582          * Returns the unique tag which must be inserted into the stripped text
 583          * The tag will be replaced with the original text in unstrip()
 584          *
 585          * @private
 586          */
 587         function insertStripItem( $text ) {
 588                 static $n = 0;
 589                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 590                 $this->mStripState->general->setPair( $rnd, $text );
 591                 return $rnd;
 592         }
 593
 594         /**
 595          * Interface with html tidy, used if $wgUseTidy = true.
 596          * If tidy isn't able to correct the markup, the original will be
 597          * returned in all its glory with a warning comment appended.
 598          *
 599          * Either the external tidy program or the in-process tidy extension
 600          * will be used depending on availability. Override the default
 601          * $wgTidyInternal setting to disable the internal if it's not working.
 602          *
 603          * @param string $text Hideous HTML input
 604          * @return string Corrected HTML output
 605          * @public
 606          * @static
 607          */
 608         function tidy( $text ) {
 609                 global $wgTidyInternal;
 610                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 611 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 612 '<head><title>test</title></head><body>'.$text.'</body></html>';
 613                 if( $wgTidyInternal ) {
 614                         $correctedtext = Parser::internalTidy( $wrappedtext );
 615                 } else {
 616                         $correctedtext = Parser::externalTidy( $wrappedtext );
 617                 }
 618                 if( is_null( $correctedtext ) ) {
 619                         wfDebug( "Tidy error detected!\n" );
 620                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 621                 }
 622                 return $correctedtext;
 623         }
 624
 625         /**
 626          * Spawn an external HTML tidy process and get corrected markup back from it.
 627          *
 628          * @private
 629          * @static
 630          */
 631         function externalTidy( $text ) {
 632                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 633                 $fname = 'Parser::externalTidy';
 634                 wfProfileIn( $fname );
 635
 636                 $cleansource = '';
 637                 $opts = ' -utf8';
 638
 639                 $descriptorspec = array(
 640                         0 => array('pipe', 'r'),
 641                         1 => array('pipe', 'w'),
 642                         2 => array('file', wfGetNull(), 'a')
 643                 );
 644                 $pipes = array();
 645                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 646                 if (is_resource($process)) {
 647                         // Theoretically, this style of communication could cause a deadlock
 648                         // here. If the stdout buffer fills up, then writes to stdin could
 649                         // block. This doesn't appear to happen with tidy, because tidy only
 650                         // writes to stdout after it's finished reading from stdin. Search
 651                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 652                         fwrite($pipes[0], $text);
 653                         fclose($pipes[0]);
 654                         while (!feof($pipes[1])) {
 655                                 $cleansource .= fgets($pipes[1], 1024);
 656                         }
 657                         fclose($pipes[1]);
 658                         proc_close($process);
 659                 }
 660
 661                 wfProfileOut( $fname );
 662
 663                 if( $cleansource == '' && $text != '') {
 664                         // Some kind of error happened, so we couldn't get the corrected text.
 665                         // Just give up; we'll use the source text and append a warning.
 666                         return null;
 667                 } else {
 668                         return $cleansource;
 669                 }
 670         }
 671
 672         /**
 673          * Use the HTML tidy PECL extension to use the tidy library in-process,
 674          * saving the overhead of spawning a new process.
 675          *
 676          * 'pear install tidy' should be able to compile the extension module.
 677          *
 678          * @private
 679          * @static
 680          */
 681         function internalTidy( $text ) {
 682                 global $wgTidyConf, $IP;
 683                 $fname = 'Parser::internalTidy';
 684                 wfProfileIn( $fname );
 685
 686                 $tidy = new tidy;
 687                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 688                 $tidy->cleanRepair();
 689                 if( $tidy->getStatus() == 2 ) {
 690                         // 2 is magic number for fatal error
 691                         // http://www.php.net/manual/en/function.tidy-get-status.php
 692                         $cleansource = null;
 693                 } else {
 694                         $cleansource = tidy_get_output( $tidy );
 695                 }
 696                 wfProfileOut( $fname );
 697                 return $cleansource;
 698         }
 699
 700         /**
 701          * parse the wiki syntax used to render tables
 702          *
 703          * @private
 704          */
 705         function doTableStuff ( $text ) {
 706                 $fname = 'Parser::doTableStuff';
 707                 wfProfileIn( $fname );
 708
 709                 $lines = explode ( "\n" , $text );
 710                 $td_history = array (); // Is currently a td tag open?
 711                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 712                 $tr_history = array (); // Is currently a tr tag open?
 713                 $tr_attributes = array (); // history of tr attributes
 714                 $has_opened_tr = array(); // Did this table open a <tr> element?
 715                 $indent_level = 0; // indent level of the table
 716                 foreach ( $lines as $key => $line )
 717                 {
 718                         $line = trim ( $line );
 719
 720                         if( $line == '' ) { // empty line, go to next line
 721                                 continue;
 722                         }
 723                         $first_character = $line{0};
 724                         $matches = array();
 725
 726                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 727                                 // First check if we are starting a new table
 728                                 $indent_level = strlen( $matches[1] );
 729
 730                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 731                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 732
 733                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 734                                 array_push ( $td_history , false );
 735                                 array_push ( $last_tag_history , '' );
 736                                 array_push ( $tr_history , false );
 737                                 array_push ( $tr_attributes , '' );
 738                                 array_push ( $has_opened_tr , false );
 739                         } else if ( count ( $td_history ) == 0 ) {
 740                                 // Don't do any of the following
 741                                 continue;
 742                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 743                                 // We are ending a table
 744                                 $line = '</table>' . substr ( $line , 2 );
 745                                 $last_tag = array_pop ( $last_tag_history );
 746
 747                                 if ( !array_pop ( $has_opened_tr ) ) {
 748                                         $line = "<tr><td></td></tr>{$line}";
 749                                 }
 750
 751                                 if ( array_pop ( $tr_history ) ) {
 752                                         $line = "</tr>{$line}";
 753                                 }
 754
 755                                 if ( array_pop ( $td_history ) ) {
 756                                         $line = "</{$last_tag}>{$line}";
 757                                 }
 758                                 array_pop ( $tr_attributes );
 759                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 760                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 761                                 // Now we have a table row
 762                                 $line = preg_replace( '#^\|-+#', '', $line );
 763
 764                                 // Whats after the tag is now only attributes
 765                                 $attributes = $this->mStripState->unstripBoth( $line );
 766                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 767                                 array_pop ( $tr_attributes );
 768                                 array_push ( $tr_attributes , $attributes );
 769
 770                                 $line = '';
 771                                 $last_tag = array_pop ( $last_tag_history );
 772                                 array_pop ( $has_opened_tr );
 773                                 array_push ( $has_opened_tr , true );
 774
 775                                 if ( array_pop ( $tr_history ) ) {
 776                                         $line = '</tr>';
 777                                 }
 778
 779                                 if ( array_pop ( $td_history ) ) {
 780                                         $line = "</{$last_tag}>{$line}";
 781                                 }
 782
 783                                 $lines[$key] = $line;
 784                                 array_push ( $tr_history , false );
 785                                 array_push ( $td_history , false );
 786                                 array_push ( $last_tag_history , '' );
 787                         }
 788                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 789                                 // This might be cell elements, td, th or captions
 790                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 791                                         $first_character = '+';
 792                                         $line = substr ( $line , 1 );
 793                                 }
 794
 795                                 $line = substr ( $line , 1 );
 796
 797                                 if ( $first_character == '!' ) {
 798                                         $line = str_replace ( '!!' , '||' , $line );
 799                                 }
 800
 801                                 // Split up multiple cells on the same line.
 802                                 // FIXME : This can result in improper nesting of tags processed
 803                                 // by earlier parser steps, but should avoid splitting up eg
 804                                 // attribute values containing literal "||".
 805                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 806
 807                                 $lines[$key] = '';
 808
 809                                 // Loop through each table cell
 810                                 foreach ( $cells as $cell )
 811                                 {
 812                                         $previous = '';
 813                                         if ( $first_character != '+' )
 814                                         {
 815                                                 $tr_after = array_pop ( $tr_attributes );
 816                                                 if ( !array_pop ( $tr_history ) ) {
 817                                                         $previous = "<tr{$tr_after}>\n";
 818                                                 }
 819                                                 array_push ( $tr_history , true );
 820                                                 array_push ( $tr_attributes , '' );
 821                                                 array_pop ( $has_opened_tr );
 822                                                 array_push ( $has_opened_tr , true );
 823                                         }
 824
 825                                         $last_tag = array_pop ( $last_tag_history );
 826
 827                                         if ( array_pop ( $td_history ) ) {
 828                                                 $previous = "</{$last_tag}>{$previous}";
 829                                         }
 830
 831                                         if ( $first_character == '|' ) {
 832                                                 $last_tag = 'td';
 833                                         } else if ( $first_character == '!' ) {
 834                                                 $last_tag = 'th';
 835                                         } else if ( $first_character == '+' ) {
 836                                                 $last_tag = 'caption';
 837                                         } else {
 838                                                 $last_tag = '';
 839                                         }
 840
 841                                         array_push ( $last_tag_history , $last_tag );
 842
 843                                         // A cell could contain both parameters and data
 844                                         $cell_data = explode ( '|' , $cell , 2 );
 845
 846                                         // Bug 553: Note that a '|' inside an invalid link should not
 847                                         // be mistaken as delimiting cell parameters
 848                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 849                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 850                                         } else if ( count ( $cell_data ) == 1 )
 851                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 852                                         else {
 853                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 854                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 855                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 856                                         }
 857
 858                                         $lines[$key] .= $cell;
 859                                         array_push ( $td_history , true );
 860                                 }
 861                         }
 862                 }
 863
 864                 // Closing open td, tr && table
 865                 while ( count ( $td_history ) > 0 )
 866                 {
 867                         if ( array_pop ( $td_history ) ) {
 868                                 $lines[] = '</td>' ;
 869                         }
 870                         if ( array_pop ( $tr_history ) ) {
 871                                 $lines[] = '</tr>' ;
 872                         }
 873                         if ( !array_pop ( $has_opened_tr ) ) {
 874                                 $lines[] = "<tr><td></td></tr>" ;
 875                         }
 876
 877                         $lines[] = '</table>' ;
 878                 }
 879
 880                 $output = implode ( "\n" , $lines ) ;
 881
 882                 // special case: don't return empty table
 883                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 884                         $output = '';
 885                 }
 886
 887                 wfProfileOut( $fname );
 888
 889                 return $output;
 890         }
 891
 892         /**
 893          * Helper function for parse() that transforms wiki markup into
 894          * HTML. Only called for $mOutputType == OT_HTML.
 895          *
 896          * @private
 897          */
 898         function internalParse( $text ) {
 899                 $isMain = true;
 900                 $fname = 'Parser::internalParse';
 901                 wfProfileIn( $fname );
 902
 903                 # Hook to suspend the parser in this state
 904                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 905                         wfProfileOut( $fname );
 906                         return $text ;
 907                 }
 908
 909                 # Remove <noinclude> tags and <includeonly> sections
 910                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 911                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 912                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 913
 914                 $text = $this->replaceVariables( $text );
 915                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 916                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 917
 918                 // Tables need to come after variable replacement for things to work
 919                 // properly; putting them before other transformations should keep
 920                 // exciting things like link expansions from showing up in surprising
 921                 // places.
 922                 $text = $this->doTableStuff( $text );
 923
 924                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 925
 926                 $text = $this->stripToc( $text );
 927                 $this->stripNoGallery( $text );
 928                 $text = $this->doHeadings( $text );
 929                 if($this->mOptions->getUseDynamicDates()) {
 930                         $df =& DateFormatter::getInstance();
 931                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 932                 }
 933                 $text = $this->doAllQuotes( $text );
 934                 $text = $this->replaceInternalLinks( $text );
 935                 $text = $this->replaceExternalLinks( $text );
 936
 937                 # replaceInternalLinks may sometimes leave behind
 938                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 939                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 940
 941                 $text = $this->doMagicLinks( $text );
 942                 $text = $this->formatHeadings( $text, $isMain );
 943
 944                 wfProfileOut( $fname );
 945                 return $text;
 946         }
 947
 948         /**
 949          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 950          * magic external links.
 951          *
 952          * @private
 953          */
 954         function doMagicLinks( $text ) {
 955                 wfProfileIn( __METHOD__ );
 956                 $text = preg_replace_callback(
 957                         '!(?:                           # Start cases
 958                             <a.*?</a> |                 # Skip link text
 959                             <.*?> |                     # Skip stuff inside HTML elements
 960                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 961                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 962                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 963                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 964                                      [0-9Xx]                 # check digit
 965                                    \b)
 966                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 967                 wfProfileOut( __METHOD__ );
 968                 return $text;
 969         }
 970
 971         function magicLinkCallback( $m ) {
 972                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 973                         # Skip HTML element
 974                         return $m[0];
 975                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 976                         $isbn = $m[2];
 977                         $num = strtr( $isbn, array(
 978                                 '-' => '',
 979                                 ' ' => '',
 980                                 'x' => 'X',
 981                         ));
 982                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
 983                         $text = '<a href="' .
 984                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
 985                                 "\" class=\"internal\">ISBN $isbn</a>";
 986                 } else {
 987                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
 988                                 $keyword = 'RFC';
 989                                 $urlmsg = 'rfcurl';
 990                                 $id = $m[1];
 991                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
 992                                 $keyword = 'PMID';
 993                                 $urlmsg = 'pubmedurl';
 994                                 $id = $m[1];
 995                         } else {
 996                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 997                                         substr($m[0], 0, 20 ) . '"' );
 998                         }
 999
1000                         $url = wfMsg( $urlmsg, $id);
1001                         $sk = $this->mOptions->getSkin();
1002                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1003                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1004                 }
1005                 return $text;
1006         }
1007
1008         /**
1009          * Parse headers and return html
1010          *
1011          * @private
1012          */
1013         function doHeadings( $text ) {
1014                 $fname = 'Parser::doHeadings';
1015                 wfProfileIn( $fname );
1016                 for ( $i = 6; $i >= 1; --$i ) {
1017                         $h = str_repeat( '=', $i );
1018                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1019                           "<h$i>\\1</h$i>", $text );
1020                 }
1021                 wfProfileOut( $fname );
1022                 return $text;
1023         }
1024
1025         /**
1026          * Replace single quotes with HTML markup
1027          * @private
1028          * @return string the altered text
1029          */
1030         function doAllQuotes( $text ) {
1031                 $fname = 'Parser::doAllQuotes';
1032                 wfProfileIn( $fname );
1033                 $outtext = '';
1034                 $lines = explode( "\n", $text );
1035                 foreach ( $lines as $line ) {
1036                         $outtext .= $this->doQuotes ( $line ) . "\n";
1037                 }
1038                 $outtext = substr($outtext, 0,-1);
1039                 wfProfileOut( $fname );
1040                 return $outtext;
1041         }
1042
1043         /**
1044          * Helper function for doAllQuotes()
1045          */
1046         public function doQuotes( $text ) {
1047                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1048                 if ( count( $arr ) == 1 )
1049                         return $text;
1050                 else
1051                 {
1052                         # First, do some preliminary work. This may shift some apostrophes from
1053                         # being mark-up to being text. It also counts the number of occurrences
1054                         # of bold and italics mark-ups.
1055                         $i = 0;
1056                         $numbold = 0;
1057                         $numitalics = 0;
1058                         foreach ( $arr as $r )
1059                         {
1060                                 if ( ( $i % 2 ) == 1 )
1061                                 {
1062                                         # If there are ever four apostrophes, assume the first is supposed to
1063                                         # be text, and the remaining three constitute mark-up for bold text.
1064                                         if ( strlen( $arr[$i] ) == 4 )
1065                                         {
1066                                                 $arr[$i-1] .= "'";
1067                                                 $arr[$i] = "'''";
1068                                         }
1069                                         # If there are more than 5 apostrophes in a row, assume they're all
1070                                         # text except for the last 5.
1071                                         else if ( strlen( $arr[$i] ) > 5 )
1072                                         {
1073                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1074                                                 $arr[$i] = "'''''";
1075                                         }
1076                                         # Count the number of occurrences of bold and italics mark-ups.
1077                                         # We are not counting sequences of five apostrophes.
1078                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1079                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1080                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1081                                 }
1082                                 $i++;
1083                         }
1084
1085                         # If there is an odd number of both bold and italics, it is likely
1086                         # that one of the bold ones was meant to be an apostrophe followed
1087                         # by italics. Which one we cannot know for certain, but it is more
1088                         # likely to be one that has a single-letter word before it.
1089                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1090                         {
1091                                 $i = 0;
1092                                 $firstsingleletterword = -1;
1093                                 $firstmultiletterword = -1;
1094                                 $firstspace = -1;
1095                                 foreach ( $arr as $r )
1096                                 {
1097                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1098                                         {
1099                                                 $x1 = substr ($arr[$i-1], -1);
1100                                                 $x2 = substr ($arr[$i-1], -2, 1);
1101                                                 if ($x1 == ' ') {
1102                                                         if ($firstspace == -1) $firstspace = $i;
1103                                                 } else if ($x2 == ' ') {
1104                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1105                                                 } else {
1106                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1107                                                 }
1108                                         }
1109                                         $i++;
1110                                 }
1111
1112                                 # If there is a single-letter word, use it!
1113                                 if ($firstsingleletterword > -1)
1114                                 {
1115                                         $arr [ $firstsingleletterword ] = "''";
1116                                         $arr [ $firstsingleletterword-1 ] .= "'";
1117                                 }
1118                                 # If not, but there's a multi-letter word, use that one.
1119                                 else if ($firstmultiletterword > -1)
1120                                 {
1121                                         $arr [ $firstmultiletterword ] = "''";
1122                                         $arr [ $firstmultiletterword-1 ] .= "'";
1123                                 }
1124                                 # ... otherwise use the first one that has neither.
1125                                 # (notice that it is possible for all three to be -1 if, for example,
1126                                 # there is only one pentuple-apostrophe in the line)
1127                                 else if ($firstspace > -1)
1128                                 {
1129                                         $arr [ $firstspace ] = "''";
1130                                         $arr [ $firstspace-1 ] .= "'";
1131                                 }
1132                         }
1133
1134                         # Now let's actually convert our apostrophic mush to HTML!
1135                         $output = '';
1136                         $buffer = '';
1137                         $state = '';
1138                         $i = 0;
1139                         foreach ($arr as $r)
1140                         {
1141                                 if (($i % 2) == 0)
1142                                 {
1143                                         if ($state == 'both')
1144                                                 $buffer .= $r;
1145                                         else
1146                                                 $output .= $r;
1147                                 }
1148                                 else
1149                                 {
1150                                         if (strlen ($r) == 2)
1151                                         {
1152                                                 if ($state == 'i')
1153                                                 { $output .= '</i>'; $state = ''; }
1154                                                 else if ($state == 'bi')
1155                                                 { $output .= '</i>'; $state = 'b'; }
1156                                                 else if ($state == 'ib')
1157                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1158                                                 else if ($state == 'both')
1159                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1160                                                 else # $state can be 'b' or ''
1161                                                 { $output .= '<i>'; $state .= 'i'; }
1162                                         }
1163                                         else if (strlen ($r) == 3)
1164                                         {
1165                                                 if ($state == 'b')
1166                                                 { $output .= '</b>'; $state = ''; }
1167                                                 else if ($state == 'bi')
1168                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1169                                                 else if ($state == 'ib')
1170                                                 { $output .= '</b>'; $state = 'i'; }
1171                                                 else if ($state == 'both')
1172                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1173                                                 else # $state can be 'i' or ''
1174                                                 { $output .= '<b>'; $state .= 'b'; }
1175                                         }
1176                                         else if (strlen ($r) == 5)
1177                                         {
1178                                                 if ($state == 'b')
1179                                                 { $output .= '</b><i>'; $state = 'i'; }
1180                                                 else if ($state == 'i')
1181                                                 { $output .= '</i><b>'; $state = 'b'; }
1182                                                 else if ($state == 'bi')
1183                                                 { $output .= '</i></b>'; $state = ''; }
1184                                                 else if ($state == 'ib')
1185                                                 { $output .= '</b></i>'; $state = ''; }
1186                                                 else if ($state == 'both')
1187                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1188                                                 else # ($state == '')
1189                                                 { $buffer = ''; $state = 'both'; }
1190                                         }
1191                                 }
1192                                 $i++;
1193                         }
1194                         # Now close all remaining tags.  Notice that the order is important.
1195                         if ($state == 'b' || $state == 'ib')
1196                                 $output .= '</b>';
1197                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1198                                 $output .= '</i>';
1199                         if ($state == 'bi')
1200                                 $output .= '</b>';
1201                         # There might be lonely ''''', so make sure we have a buffer
1202                         if ($state == 'both' && $buffer)
1203                                 $output .= '<b><i>'.$buffer.'</i></b>';
1204                         return $output;
1205                 }
1206         }
1207
1208         /**
1209          * Replace external links
1210          *
1211          * Note: this is all very hackish and the order of execution matters a lot.
1212          * Make sure to run maintenance/parserTests.php if you change this code.
1213          *
1214          * @private
1215          */
1216         function replaceExternalLinks( $text ) {
1217                 global $wgContLang;
1218                 $fname = 'Parser::replaceExternalLinks';
1219                 wfProfileIn( $fname );
1220
1221                 $sk = $this->mOptions->getSkin();
1222
1223                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1224
1225                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1226
1227                 $i = 0;
1228                 while ( $i<count( $bits ) ) {
1229                         $url = $bits[$i++];
1230                         $protocol = $bits[$i++];
1231                         $text = $bits[$i++];
1232                         $trail = $bits[$i++];
1233
1234                         # The characters '<' and '>' (which were escaped by
1235                         # removeHTMLtags()) should not be included in
1236                         # URLs, per RFC 2396.
1237                         $m2 = array();
1238                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1239                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1240                                 $url = substr($url, 0, $m2[0][1]);
1241                         }
1242
1243                         # If the link text is an image URL, replace it with an <img> tag
1244                         # This happened by accident in the original parser, but some people used it extensively
1245                         $img = $this->maybeMakeExternalImage( $text );
1246                         if ( $img !== false ) {
1247                                 $text = $img;
1248                         }
1249
1250                         $dtrail = '';
1251
1252                         # Set linktype for CSS - if URL==text, link is essentially free
1253                         $linktype = ($text == $url) ? 'free' : 'text';
1254
1255                         # No link text, e.g. [http://domain.tld/some.link]
1256                         if ( $text == '' ) {
1257                                 # Autonumber if allowed. See bug #5918
1258                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1259                                         $text = '[' . ++$this->mAutonumber . ']';
1260                                         $linktype = 'autonumber';
1261                                 } else {
1262                                         # Otherwise just use the URL
1263                                         $text = htmlspecialchars( $url );
1264                                         $linktype = 'free';
1265                                 }
1266                         } else {
1267                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1268                                 # Check for trail
1269                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1270                         }
1271
1272                         $text = $wgContLang->markNoConversion($text);
1273
1274                         $url = Sanitizer::cleanUrl( $url );
1275
1276                         # Process the trail (i.e. everything after this link up until start of the next link),
1277                         # replacing any non-bracketed links
1278                         $trail = $this->replaceFreeExternalLinks( $trail );
1279
1280                         # Use the encoded URL
1281                         # This means that users can paste URLs directly into the text
1282                         # Funny characters like &ouml; aren't valid in URLs anyway
1283                         # This was changed in August 2004
1284                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1285
1286                         # Register link in the output object.
1287                         # Replace unnecessary URL escape codes with the referenced character
1288                         # This prevents spammers from hiding links from the filters
1289                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1290                         $this->mOutput->addExternalLink( $pasteurized );
1291                 }
1292
1293                 wfProfileOut( $fname );
1294                 return $s;
1295         }
1296
1297         /**
1298          * Replace anything that looks like a URL with a link
1299          * @private
1300          */
1301         function replaceFreeExternalLinks( $text ) {
1302                 global $wgContLang;
1303                 $fname = 'Parser::replaceFreeExternalLinks';
1304                 wfProfileIn( $fname );
1305
1306                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1307                 $s = array_shift( $bits );
1308                 $i = 0;
1309
1310                 $sk = $this->mOptions->getSkin();
1311
1312                 while ( $i < count( $bits ) ){
1313                         $protocol = $bits[$i++];
1314                         $remainder = $bits[$i++];
1315
1316                         $m = array();
1317                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1318                                 # Found some characters after the protocol that look promising
1319                                 $url = $protocol . $m[1];
1320                                 $trail = $m[2];
1321
1322                                 # special case: handle urls as url args:
1323                                 # http://www.example.com/foo?=http://www.example.com/bar
1324                                 if(strlen($trail) == 0 &&
1325                                         isset($bits[$i]) &&
1326                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1327                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1328                                 {
1329                                         # add protocol, arg
1330                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1331                                         $i += 2;
1332                                         $trail = $m[2];
1333                                 }
1334
1335                                 # The characters '<' and '>' (which were escaped by
1336                                 # removeHTMLtags()) should not be included in
1337                                 # URLs, per RFC 2396.
1338                                 $m2 = array();
1339                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1340                                         $trail = substr($url, $m2[0][1]) . $trail;
1341                                         $url = substr($url, 0, $m2[0][1]);
1342                                 }
1343
1344                                 # Move trailing punctuation to $trail
1345                                 $sep = ',;\.:!?';
1346                                 # If there is no left bracket, then consider right brackets fair game too
1347                                 if ( strpos( $url, '(' ) === false ) {
1348                                         $sep .= ')';
1349                                 }
1350
1351                                 $numSepChars = strspn( strrev( $url ), $sep );
1352                                 if ( $numSepChars ) {
1353                                         $trail = substr( $url, -$numSepChars ) . $trail;
1354                                         $url = substr( $url, 0, -$numSepChars );
1355                                 }
1356
1357                                 $url = Sanitizer::cleanUrl( $url );
1358
1359                                 # Is this an external image?
1360                                 $text = $this->maybeMakeExternalImage( $url );
1361                                 if ( $text === false ) {
1362                                         # Not an image, make a link
1363                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1364                                         # Register it in the output object...
1365                                         # Replace unnecessary URL escape codes with their equivalent characters
1366                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1367                                         $this->mOutput->addExternalLink( $pasteurized );
1368                                 }
1369                                 $s .= $text . $trail;
1370                         } else {
1371                                 $s .= $protocol . $remainder;
1372                         }
1373                 }
1374                 wfProfileOut( $fname );
1375                 return $s;
1376         }
1377
1378         /**
1379          * Replace unusual URL escape codes with their equivalent characters
1380          * @param string
1381          * @return string
1382          * @static
1383          * @todo  This can merge genuinely required bits in the path or query string,
1384          *        breaking legit URLs. A proper fix would treat the various parts of
1385          *        the URL differently; as a workaround, just use the output for
1386          *        statistical records, not for actual linking/output.
1387          */
1388         static function replaceUnusualEscapes( $url ) {
1389                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1390                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1391         }
1392
1393         /**
1394          * Callback function used in replaceUnusualEscapes().
1395          * Replaces unusual URL escape codes with their equivalent character
1396          * @static
1397          * @private
1398          */
1399         private static function replaceUnusualEscapesCallback( $matches ) {
1400                 $char = urldecode( $matches[0] );
1401                 $ord = ord( $char );
1402                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1403                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1404                         // No, shouldn't be escaped
1405                         return $char;
1406                 } else {
1407                         // Yes, leave it escaped
1408                         return $matches[0];
1409                 }
1410         }
1411
1412         /**
1413          * make an image if it's allowed, either through the global
1414          * option or through the exception
1415          * @private
1416          */
1417         function maybeMakeExternalImage( $url ) {
1418                 $sk = $this->mOptions->getSkin();
1419                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1420                 $imagesexception = !empty($imagesfrom);
1421                 $text = false;
1422                 if ( $this->mOptions->getAllowExternalImages()
1423                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1424                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1425                                 # Image found
1426                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1427                         }
1428                 }
1429                 return $text;
1430         }
1431
1432         /**
1433          * Process [[ ]] wikilinks
1434          *
1435          * @private
1436          */
1437         function replaceInternalLinks( $s ) {
1438                 global $wgContLang;
1439                 static $fname = 'Parser::replaceInternalLinks' ;
1440
1441                 wfProfileIn( $fname );
1442
1443                 wfProfileIn( $fname.'-setup' );
1444                 static $tc = FALSE;
1445                 # the % is needed to support urlencoded titles as well
1446                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1447
1448                 $sk = $this->mOptions->getSkin();
1449
1450                 #split the entire text string on occurences of [[
1451                 $a = explode( '[[', ' ' . $s );
1452                 #get the first element (all text up to first [[), and remove the space we added
1453                 $s = array_shift( $a );
1454                 $s = substr( $s, 1 );
1455
1456                 # Match a link having the form [[namespace:link|alternate]]trail
1457                 static $e1 = FALSE;
1458                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1459                 # Match cases where there is no "]]", which might still be images
1460                 static $e1_img = FALSE;
1461                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1462
1463                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1464                 $e2 = null;
1465                 if ( $useLinkPrefixExtension ) {
1466                         # Match the end of a line for a word that's not followed by whitespace,
1467                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1468                         $e2 = wfMsgForContent( 'linkprefix' );
1469                 }
1470
1471                 if( is_null( $this->mTitle ) ) {
1472                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1473                 }
1474                 $nottalk = !$this->mTitle->isTalkPage();
1475
1476                 if ( $useLinkPrefixExtension ) {
1477                         $m = array();
1478                         if ( preg_match( $e2, $s, $m ) ) {
1479                                 $first_prefix = $m[2];
1480                         } else {
1481                                 $first_prefix = false;
1482                         }
1483                 } else {
1484                         $prefix = '';
1485                 }
1486
1487                 if($wgContLang->hasVariants()) {
1488                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1489                 } else {
1490                         $selflink = array($this->mTitle->getPrefixedText());
1491                 }
1492                 $useSubpages = $this->areSubpagesAllowed();
1493                 wfProfileOut( $fname.'-setup' );
1494
1495                 # Loop for each link
1496                 for ($k = 0; isset( $a[$k] ); $k++) {
1497                         $line = $a[$k];
1498                         if ( $useLinkPrefixExtension ) {
1499                                 wfProfileIn( $fname.'-prefixhandling' );
1500                                 if ( preg_match( $e2, $s, $m ) ) {
1501                                         $prefix = $m[2];
1502                                         $s = $m[1];
1503                                 } else {
1504                                         $prefix='';
1505                                 }
1506                                 # first link
1507                                 if($first_prefix) {
1508                                         $prefix = $first_prefix;
1509                                         $first_prefix = false;
1510                                 }
1511                                 wfProfileOut( $fname.'-prefixhandling' );
1512                         }
1513
1514                         $might_be_img = false;
1515
1516                         wfProfileIn( "$fname-e1" );
1517                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1518                                 $text = $m[2];
1519                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1520                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1521                                 # the real problem is with the $e1 regex
1522                                 # See bug 1300.
1523                                 #
1524                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1525                                 # and no image is in sight. See bug 2095.
1526                                 #
1527                                 if( $text !== '' &&
1528                                         substr( $m[3], 0, 1 ) === ']' &&
1529                                         strpos($text, '[') !== false
1530                                 )
1531                                 {
1532                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1533                                         $m[3] = substr( $m[3], 1 );
1534                                 }
1535                                 # fix up urlencoded title texts
1536                                 if( strpos( $m[1], '%' ) !== false ) {
1537                                         # Should anchors '#' also be rejected?
1538                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1539                                 }
1540                                 $trail = $m[3];
1541                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1542                                 $might_be_img = true;
1543                                 $text = $m[2];
1544                                 if ( strpos( $m[1], '%' ) !== false ) {
1545                                         $m[1] = urldecode($m[1]);
1546                                 }
1547                                 $trail = "";
1548                         } else { # Invalid form; output directly
1549                                 $s .= $prefix . '[[' . $line ;
1550                                 wfProfileOut( "$fname-e1" );
1551                                 continue;
1552                         }
1553                         wfProfileOut( "$fname-e1" );
1554                         wfProfileIn( "$fname-misc" );
1555
1556                         # Don't allow internal links to pages containing
1557                         # PROTO: where PROTO is a valid URL protocol; these
1558                         # should be external links.
1559                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1560                                 $s .= $prefix . '[[' . $line ;
1561                                 continue;
1562                         }
1563
1564                         # Make subpage if necessary
1565                         if( $useSubpages ) {
1566                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1567                         } else {
1568                                 $link = $m[1];
1569                         }
1570
1571                         $noforce = (substr($m[1], 0, 1) != ':');
1572                         if (!$noforce) {
1573                                 # Strip off leading ':'
1574                                 $link = substr($link, 1);
1575                         }
1576
1577                         wfProfileOut( "$fname-misc" );
1578                         wfProfileIn( "$fname-title" );
1579                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1580                         if( !$nt ) {
1581                                 $s .= $prefix . '[[' . $line;
1582                                 wfProfileOut( "$fname-title" );
1583                                 continue;
1584                         }
1585
1586                         $ns = $nt->getNamespace();
1587                         $iw = $nt->getInterWiki();
1588                         wfProfileOut( "$fname-title" );
1589
1590                         if ($might_be_img) { # if this is actually an invalid link
1591                                 wfProfileIn( "$fname-might_be_img" );
1592                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1593                                         $found = false;
1594                                         while (isset ($a[$k+1]) ) {
1595                                                 #look at the next 'line' to see if we can close it there
1596                                                 $spliced = array_splice( $a, $k + 1, 1 );
1597                                                 $next_line = array_shift( $spliced );
1598                                                 $m = explode( ']]', $next_line, 3 );
1599                                                 if ( count( $m ) == 3 ) {
1600                                                         # the first ]] closes the inner link, the second the image
1601                                                         $found = true;
1602                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1603                                                         $trail = $m[2];
1604                                                         break;
1605                                                 } elseif ( count( $m ) == 2 ) {
1606                                                         #if there's exactly one ]] that's fine, we'll keep looking
1607                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1608                                                 } else {
1609                                                         #if $next_line is invalid too, we need look no further
1610                                                         $text .= '[[' . $next_line;
1611                                                         break;
1612                                                 }
1613                                         }
1614                                         if ( !$found ) {
1615                                                 # we couldn't find the end of this imageLink, so output it raw
1616                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1617                                                 $text = $this->replaceInternalLinks($text);
1618                                                 $s .= "{$prefix}[[$link|$text";
1619                                                 # note: no $trail, because without an end, there *is* no trail
1620                                                 wfProfileOut( "$fname-might_be_img" );
1621                                                 continue;
1622                                         }
1623                                 } else { #it's not an image, so output it raw
1624                                         $s .= "{$prefix}[[$link|$text";
1625                                         # note: no $trail, because without an end, there *is* no trail
1626                                         wfProfileOut( "$fname-might_be_img" );
1627                                         continue;
1628                                 }
1629                                 wfProfileOut( "$fname-might_be_img" );
1630                         }
1631
1632                         $wasblank = ( '' == $text );
1633                         if( $wasblank ) $text = $link;
1634
1635                         # Link not escaped by : , create the various objects
1636                         if( $noforce ) {
1637
1638                                 # Interwikis
1639                                 wfProfileIn( "$fname-interwiki" );
1640                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1641                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1642                                         $s = rtrim($s . $prefix);
1643                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1644                                         wfProfileOut( "$fname-interwiki" );
1645                                         continue;
1646                                 }
1647                                 wfProfileOut( "$fname-interwiki" );
1648
1649                                 if ( $ns == NS_IMAGE ) {
1650                                         wfProfileIn( "$fname-image" );
1651                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1652                                                 # recursively parse links inside the image caption
1653                                                 # actually, this will parse them in any other parameters, too,
1654                                                 # but it might be hard to fix that, and it doesn't matter ATM
1655                                                 $text = $this->replaceExternalLinks($text);
1656                                                 $text = $this->replaceInternalLinks($text);
1657
1658                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1659                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1660                                                 $this->mOutput->addImage( $nt->getDBkey() );
1661
1662                                                 wfProfileOut( "$fname-image" );
1663                                                 continue;
1664                                         } else {
1665                                                 # We still need to record the image's presence on the page
1666                                                 $this->mOutput->addImage( $nt->getDBkey() );
1667                                         }
1668                                         wfProfileOut( "$fname-image" );
1669
1670                                 }
1671
1672                                 if ( $ns == NS_CATEGORY ) {
1673                                         wfProfileIn( "$fname-category" );
1674                                         $s = rtrim($s . "\n"); # bug 87
1675
1676                                         if ( $wasblank ) {
1677                                                 $sortkey = $this->getDefaultSort();
1678                                         } else {
1679                                                 $sortkey = $text;
1680                                         }
1681                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1682                                         $sortkey = str_replace( "\n", '', $sortkey );
1683                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1684                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1685
1686                                         /**
1687                                          * Strip the whitespace Category links produce, see bug 87
1688                                          * @todo We might want to use trim($tmp, "\n") here.
1689                                          */
1690                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1691
1692                                         wfProfileOut( "$fname-category" );
1693                                         continue;
1694                                 }
1695                         }
1696
1697                         # Self-link checking
1698                         if( $nt->getFragment() === '' ) {
1699                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1700                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1701                                         continue;
1702                                 }
1703                         }
1704
1705                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1706                         if( $ns == NS_MEDIA ) {
1707                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1708                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1709                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1710                                 $this->mOutput->addImage( $nt->getDBkey() );
1711                                 continue;
1712                         } elseif( $ns == NS_SPECIAL ) {
1713                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1714                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1715                                 } else {
1716                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1717                                 }
1718                                 continue;
1719                         } elseif( $ns == NS_IMAGE ) {
1720                                 $img = wfFindFile( $nt );
1721                                 if( $img ) {
1722                                         // Force a blue link if the file exists; may be a remote
1723                                         // upload on the shared repository, and we want to see its
1724                                         // auto-generated page.
1725                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1726                                         $this->mOutput->addLink( $nt );
1727                                         continue;
1728                                 }
1729                         }
1730                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1731                 }
1732                 wfProfileOut( $fname );
1733                 return $s;
1734         }
1735
1736         /**
1737          * Make a link placeholder. The text returned can be later resolved to a real link with
1738          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1739          * parsing of interwiki links, and secondly to allow all existence checks and
1740          * article length checks (for stub links) to be bundled into a single query.
1741          *
1742          */
1743         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1744                 wfProfileIn( __METHOD__ );
1745                 if ( ! is_object($nt) ) {
1746                         # Fail gracefully
1747                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1748                 } else {
1749                         # Separate the link trail from the rest of the link
1750                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1751
1752                         if ( $nt->isExternal() ) {
1753                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1754                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1755                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1756                         } else {
1757                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1758                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1759                                 $this->mLinkHolders['queries'][] = $query;
1760                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1761                                 $this->mLinkHolders['titles'][] = $nt;
1762
1763                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1764                         }
1765                 }
1766                 wfProfileOut( __METHOD__ );
1767                 return $retVal;
1768         }
1769
1770         /**
1771          * Render a forced-blue link inline; protect against double expansion of
1772          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1773          * Since this little disaster has to split off the trail text to avoid
1774          * breaking URLs in the following text without breaking trails on the
1775          * wiki links, it's been made into a horrible function.
1776          *
1777          * @param Title $nt
1778          * @param string $text
1779          * @param string $query
1780          * @param string $trail
1781          * @param string $prefix
1782          * @return string HTML-wikitext mix oh yuck
1783          */
1784         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1785                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1786                 $sk = $this->mOptions->getSkin();
1787                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1788                 return $this->armorLinks( $link ) . $trail;
1789         }
1790
1791         /**
1792          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1793          * going to go through further parsing steps before inline URL expansion.
1794          *
1795          * In particular this is important when using action=render, which causes
1796          * full URLs to be included.
1797          *
1798          * Oh man I hate our multi-layer parser!
1799          *
1800          * @param string more-or-less HTML
1801          * @return string less-or-more HTML with NOPARSE bits
1802          */
1803         function armorLinks( $text ) {
1804                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1805                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1806         }
1807
1808         /**
1809          * Return true if subpage links should be expanded on this page.
1810          * @return bool
1811          */
1812         function areSubpagesAllowed() {
1813                 # Some namespaces don't allow subpages
1814                 global $wgNamespacesWithSubpages;
1815                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1816         }
1817
1818         /**
1819          * Handle link to subpage if necessary
1820          * @param string $target the source of the link
1821          * @param string &$text the link text, modified as necessary
1822          * @return string the full name of the link
1823          * @private
1824          */
1825         function maybeDoSubpageLink($target, &$text) {
1826                 # Valid link forms:
1827                 # Foobar -- normal
1828                 # :Foobar -- override special treatment of prefix (images, language links)
1829                 # /Foobar -- convert to CurrentPage/Foobar
1830                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1831                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1832                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1833
1834                 $fname = 'Parser::maybeDoSubpageLink';
1835                 wfProfileIn( $fname );
1836                 $ret = $target; # default return value is no change
1837
1838                 # Some namespaces don't allow subpages,
1839                 # so only perform processing if subpages are allowed
1840                 if( $this->areSubpagesAllowed() ) {
1841                         $hash = strpos( $target, '#' );
1842                         if( $hash !== false ) {
1843                                 $suffix = substr( $target, $hash );
1844                                 $target = substr( $target, 0, $hash );
1845                         } else {
1846                                 $suffix = '';
1847                         }
1848                         # bug 7425
1849                         $target = trim( $target );
1850                         # Look at the first character
1851                         if( $target != '' && $target{0} == '/' ) {
1852                                 # / at end means we don't want the slash to be shown
1853                                 $m = array();
1854                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1855                                 if( $trailingSlashes ) {
1856                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1857                                 } else {
1858                                         $noslash = substr( $target, 1 );
1859                                 }
1860
1861                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1862                                 if( '' === $text ) {
1863                                         $text = $target . $suffix;
1864                                 } # this might be changed for ugliness reasons
1865                         } else {
1866                                 # check for .. subpage backlinks
1867                                 $dotdotcount = 0;
1868                                 $nodotdot = $target;
1869                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1870                                         ++$dotdotcount;
1871                                         $nodotdot = substr( $nodotdot, 3 );
1872                                 }
1873                                 if($dotdotcount > 0) {
1874                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1875                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1876                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1877                                                 # / at the end means don't show full path
1878                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1879                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1880                                                         if( '' === $text ) {
1881                                                                 $text = $nodotdot . $suffix;
1882                                                         }
1883                                                 }
1884                                                 $nodotdot = trim( $nodotdot );
1885                                                 if( $nodotdot != '' ) {
1886                                                         $ret .= '/' . $nodotdot;
1887                                                 }
1888                                                 $ret .= $suffix;
1889                                         }
1890                                 }
1891                         }
1892                 }
1893
1894                 wfProfileOut( $fname );
1895                 return $ret;
1896         }
1897
1898         /**#@+
1899          * Used by doBlockLevels()
1900          * @private
1901          */
1902         /* private */ function closeParagraph() {
1903                 $result = '';
1904                 if ( '' != $this->mLastSection ) {
1905                         $result = '</' . $this->mLastSection  . ">\n";
1906                 }
1907                 $this->mInPre = false;
1908                 $this->mLastSection = '';
1909                 return $result;
1910         }
1911         # getCommon() returns the length of the longest common substring
1912         # of both arguments, starting at the beginning of both.
1913         #
1914         /* private */ function getCommon( $st1, $st2 ) {
1915                 $fl = strlen( $st1 );
1916                 $shorter = strlen( $st2 );
1917                 if ( $fl < $shorter ) { $shorter = $fl; }
1918
1919                 for ( $i = 0; $i < $shorter; ++$i ) {
1920                         if ( $st1{$i} != $st2{$i} ) { break; }
1921                 }
1922                 return $i;
1923         }
1924         # These next three functions open, continue, and close the list
1925         # element appropriate to the prefix character passed into them.
1926         #
1927         /* private */ function openList( $char ) {
1928                 $result = $this->closeParagraph();
1929
1930                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1931                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1932                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1933                 else if ( ';' == $char ) {
1934                         $result .= '<dl><dt>';
1935                         $this->mDTopen = true;
1936                 }
1937                 else { $result = '<!-- ERR 1 -->'; }
1938
1939                 return $result;
1940         }
1941
1942         /* private */ function nextItem( $char ) {
1943                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1944                 else if ( ':' == $char || ';' == $char ) {
1945                         $close = '</dd>';
1946                         if ( $this->mDTopen ) { $close = '</dt>'; }
1947                         if ( ';' == $char ) {
1948                                 $this->mDTopen = true;
1949                                 return $close . '<dt>';
1950                         } else {
1951                                 $this->mDTopen = false;
1952                                 return $close . '<dd>';
1953                         }
1954                 }
1955                 return '<!-- ERR 2 -->';
1956         }
1957
1958         /* private */ function closeList( $char ) {
1959                 if ( '*' == $char ) { $text = '</li></ul>'; }
1960                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1961                 else if ( ':' == $char ) {
1962                         if ( $this->mDTopen ) {
1963                                 $this->mDTopen = false;
1964                                 $text = '</dt></dl>';
1965                         } else {
1966                                 $text = '</dd></dl>';
1967                         }
1968                 }
1969                 else {  return '<!-- ERR 3 -->'; }
1970                 return $text."\n";
1971         }
1972         /**#@-*/
1973
1974         /**
1975          * Make lists from lines starting with ':', '*', '#', etc.
1976          *
1977          * @private
1978          * @return string the lists rendered as HTML
1979          */
1980         function doBlockLevels( $text, $linestart ) {
1981                 $fname = 'Parser::doBlockLevels';
1982                 wfProfileIn( $fname );
1983
1984                 # Parsing through the text line by line.  The main thing
1985                 # happening here is handling of block-level elements p, pre,
1986                 # and making lists from lines starting with * # : etc.
1987                 #
1988                 $textLines = explode( "\n", $text );
1989
1990                 $lastPrefix = $output = '';
1991                 $this->mDTopen = $inBlockElem = false;
1992                 $prefixLength = 0;
1993                 $paragraphStack = false;
1994
1995                 if ( !$linestart ) {
1996                         $output .= array_shift( $textLines );
1997                 }
1998                 foreach ( $textLines as $oLine ) {
1999                         $lastPrefixLength = strlen( $lastPrefix );
2000                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2001                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2002                         if ( !$this->mInPre ) {
2003                                 # Multiple prefixes may abut each other for nested lists.
2004                                 $prefixLength = strspn( $oLine, '*#:;' );
2005                                 $pref = substr( $oLine, 0, $prefixLength );
2006
2007                                 # eh?
2008                                 $pref2 = str_replace( ';', ':', $pref );
2009                                 $t = substr( $oLine, $prefixLength );
2010                                 $this->mInPre = !empty($preOpenMatch);
2011                         } else {
2012                                 # Don't interpret any other prefixes in preformatted text
2013                                 $prefixLength = 0;
2014                                 $pref = $pref2 = '';
2015                                 $t = $oLine;
2016                         }
2017
2018                         # List generation
2019                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2020                                 # Same as the last item, so no need to deal with nesting or opening stuff
2021                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2022                                 $paragraphStack = false;
2023
2024                                 if ( substr( $pref, -1 ) == ';') {
2025                                         # The one nasty exception: definition lists work like this:
2026                                         # ; title : definition text
2027                                         # So we check for : in the remainder text to split up the
2028                                         # title and definition, without b0rking links.
2029                                         $term = $t2 = '';
2030                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2031                                                 $t = $t2;
2032                                                 $output .= $term . $this->nextItem( ':' );
2033                                         }
2034                                 }
2035                         } elseif( $prefixLength || $lastPrefixLength ) {
2036                                 # Either open or close a level...
2037                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2038                                 $paragraphStack = false;
2039
2040                                 while( $commonPrefixLength < $lastPrefixLength ) {
2041                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2042                                         --$lastPrefixLength;
2043                                 }
2044                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2045                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2046                                 }
2047                                 while ( $prefixLength > $commonPrefixLength ) {
2048                                         $char = substr( $pref, $commonPrefixLength, 1 );
2049                                         $output .= $this->openList( $char );
2050
2051                                         if ( ';' == $char ) {
2052                                                 # FIXME: This is dupe of code above
2053                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2054                                                         $t = $t2;
2055                                                         $output .= $term . $this->nextItem( ':' );
2056                                                 }
2057                                         }
2058                                         ++$commonPrefixLength;
2059                                 }
2060                                 $lastPrefix = $pref2;
2061                         }
2062                         if( 0 == $prefixLength ) {
2063                                 wfProfileIn( "$fname-paragraph" );
2064                                 # No prefix (not in list)--go to paragraph mode
2065                                 // XXX: use a stack for nestable elements like span, table and div
2066                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2067                                 $closematch = preg_match(
2068                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2069                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2070                                 if ( $openmatch or $closematch ) {
2071                                         $paragraphStack = false;
2072                                         # TODO bug 5718: paragraph closed
2073                                         $output .= $this->closeParagraph();
2074                                         if ( $preOpenMatch and !$preCloseMatch ) {
2075                                                 $this->mInPre = true;
2076                                         }
2077                                         if ( $closematch ) {
2078                                                 $inBlockElem = false;
2079                                         } else {
2080                                                 $inBlockElem = true;
2081                                         }
2082                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2083                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2084                                                 // pre
2085                                                 if ($this->mLastSection != 'pre') {
2086                                                         $paragraphStack = false;
2087                                                         $output .= $this->closeParagraph().'<pre>';
2088                                                         $this->mLastSection = 'pre';
2089                                                 }
2090                                                 $t = substr( $t, 1 );
2091                                         } else {
2092                                                 // paragraph
2093                                                 if ( '' == trim($t) ) {
2094                                                         if ( $paragraphStack ) {
2095                                                                 $output .= $paragraphStack.'<br />';
2096                                                                 $paragraphStack = false;
2097                                                                 $this->mLastSection = 'p';
2098                                                         } else {
2099                                                                 if ($this->mLastSection != 'p' ) {
2100                                                                         $output .= $this->closeParagraph();
2101                                                                         $this->mLastSection = '';
2102                                                                         $paragraphStack = '<p>';
2103                                                                 } else {
2104                                                                         $paragraphStack = '</p><p>';
2105                                                                 }
2106                                                         }
2107                                                 } else {
2108                                                         if ( $paragraphStack ) {
2109                                                                 $output .= $paragraphStack;
2110                                                                 $paragraphStack = false;
2111                                                                 $this->mLastSection = 'p';
2112                                                         } else if ($this->mLastSection != 'p') {
2113                                                                 $output .= $this->closeParagraph().'<p>';
2114                                                                 $this->mLastSection = 'p';
2115                                                         }
2116                                                 }
2117                                         }
2118                                 }
2119                                 wfProfileOut( "$fname-paragraph" );
2120                         }
2121                         // somewhere above we forget to get out of pre block (bug 785)
2122                         if($preCloseMatch && $this->mInPre) {
2123                                 $this->mInPre = false;
2124                         }
2125                         if ($paragraphStack === false) {
2126                                 $output .= $t."\n";
2127                         }
2128                 }
2129                 while ( $prefixLength ) {
2130                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2131                         --$prefixLength;
2132                 }
2133                 if ( '' != $this->mLastSection ) {
2134                         $output .= '</' . $this->mLastSection . '>';
2135                         $this->mLastSection = '';
2136                 }
2137
2138                 wfProfileOut( $fname );
2139                 return $output;
2140         }
2141
2142         /**
2143          * Split up a string on ':', ignoring any occurences inside tags
2144          * to prevent illegal overlapping.
2145          * @param string $str the string to split
2146          * @param string &$before set to everything before the ':'
2147          * @param string &$after set to everything after the ':'
2148          * return string the position of the ':', or false if none found
2149          */
2150         function findColonNoLinks($str, &$before, &$after) {
2151                 $fname = 'Parser::findColonNoLinks';
2152                 wfProfileIn( $fname );
2153
2154                 $pos = strpos( $str, ':' );
2155                 if( $pos === false ) {
2156                         // Nothing to find!
2157                         wfProfileOut( $fname );
2158                         return false;
2159                 }
2160
2161                 $lt = strpos( $str, '<' );
2162                 if( $lt === false || $lt > $pos ) {
2163                         // Easy; no tag nesting to worry about
2164                         $before = substr( $str, 0, $pos );
2165                         $after = substr( $str, $pos+1 );
2166                         wfProfileOut( $fname );
2167                         return $pos;
2168                 }
2169
2170                 // Ugly state machine to walk through avoiding tags.
2171                 $state = self::COLON_STATE_TEXT;
2172                 $stack = 0;
2173                 $len = strlen( $str );
2174                 for( $i = 0; $i < $len; $i++ ) {
2175                         $c = $str{$i};
2176
2177                         switch( $state ) {
2178                         // (Using the number is a performance hack for common cases)
2179                         case 0: // self::COLON_STATE_TEXT:
2180                                 switch( $c ) {
2181                                 case "<":
2182                                         // Could be either a <start> tag or an </end> tag
2183                                         $state = self::COLON_STATE_TAGSTART;
2184                                         break;
2185                                 case ":":
2186                                         if( $stack == 0 ) {
2187                                                 // We found it!
2188                                                 $before = substr( $str, 0, $i );
2189                                                 $after = substr( $str, $i + 1 );
2190                                                 wfProfileOut( $fname );
2191                                                 return $i;
2192                                         }
2193                                         // Embedded in a tag; don't break it.
2194                                         break;
2195                                 default:
2196                                         // Skip ahead looking for something interesting
2197                                         $colon = strpos( $str, ':', $i );
2198                                         if( $colon === false ) {
2199                                                 // Nothing else interesting
2200                                                 wfProfileOut( $fname );
2201                                                 return false;
2202                                         }
2203                                         $lt = strpos( $str, '<', $i );
2204                                         if( $stack === 0 ) {
2205                                                 if( $lt === false || $colon < $lt ) {
2206                                                         // We found it!
2207                                                         $before = substr( $str, 0, $colon );
2208                                                         $after = substr( $str, $colon + 1 );
2209                                                         wfProfileOut( $fname );
2210                                                         return $i;
2211                                                 }
2212                                         }
2213                                         if( $lt === false ) {
2214                                                 // Nothing else interesting to find; abort!
2215                                                 // We're nested, but there's no close tags left. Abort!
2216                                                 break 2;
2217                                         }
2218                                         // Skip ahead to next tag start
2219                                         $i = $lt;
2220                                         $state = self::COLON_STATE_TAGSTART;
2221                                 }
2222                                 break;
2223                         case 1: // self::COLON_STATE_TAG:
2224                                 // In a <tag>
2225                                 switch( $c ) {
2226                                 case ">":
2227                                         $stack++;
2228                                         $state = self::COLON_STATE_TEXT;
2229                                         break;
2230                                 case "/":
2231                                         // Slash may be followed by >?
2232                                         $state = self::COLON_STATE_TAGSLASH;
2233                                         break;
2234                                 default:
2235                                         // ignore
2236                                 }
2237                                 break;
2238                         case 2: // self::COLON_STATE_TAGSTART:
2239                                 switch( $c ) {
2240                                 case "/":
2241                                         $state = self::COLON_STATE_CLOSETAG;
2242                                         break;
2243                                 case "!":
2244                                         $state = self::COLON_STATE_COMMENT;
2245                                         break;
2246                                 case ">":
2247                                         // Illegal early close? This shouldn't happen D:
2248                                         $state = self::COLON_STATE_TEXT;
2249                                         break;
2250                                 default:
2251                                         $state = self::COLON_STATE_TAG;
2252                                 }
2253                                 break;
2254                         case 3: // self::COLON_STATE_CLOSETAG:
2255                                 // In a </tag>
2256                                 if( $c == ">" ) {
2257                                         $stack--;
2258                                         if( $stack < 0 ) {
2259                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2260                                                 wfProfileOut( $fname );
2261                                                 return false;
2262                                         }
2263                                         $state = self::COLON_STATE_TEXT;
2264                                 }
2265                                 break;
2266                         case self::COLON_STATE_TAGSLASH:
2267                                 if( $c == ">" ) {
2268                                         // Yes, a self-closed tag <blah/>
2269                                         $state = self::COLON_STATE_TEXT;
2270                                 } else {
2271                                         // Probably we're jumping the gun, and this is an attribute
2272                                         $state = self::COLON_STATE_TAG;
2273                                 }
2274                                 break;
2275                         case 5: // self::COLON_STATE_COMMENT:
2276                                 if( $c == "-" ) {
2277                                         $state = self::COLON_STATE_COMMENTDASH;
2278                                 }
2279                                 break;
2280                         case self::COLON_STATE_COMMENTDASH:
2281                                 if( $c == "-" ) {
2282                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2283                                 } else {
2284                                         $state = self::COLON_STATE_COMMENT;
2285                                 }
2286                                 break;
2287                         case self::COLON_STATE_COMMENTDASHDASH:
2288                                 if( $c == ">" ) {
2289                                         $state = self::COLON_STATE_TEXT;
2290                                 } else {
2291                                         $state = self::COLON_STATE_COMMENT;
2292                                 }
2293                                 break;
2294                         default:
2295                                 throw new MWException( "State machine error in $fname" );
2296                         }
2297                 }
2298                 if( $stack > 0 ) {
2299                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2300                         return false;
2301                 }
2302                 wfProfileOut( $fname );
2303                 return false;
2304         }
2305
2306         /**
2307          * Return value of a magic variable (like PAGENAME)
2308          *
2309          * @private
2310          */
2311         function getVariableValue( $index ) {
2312                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2313
2314                 /**
2315                  * Some of these require message or data lookups and can be
2316                  * expensive to check many times.
2317                  */
2318                 static $varCache = array();
2319                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2320                         if ( isset( $varCache[$index] ) ) {
2321                                 return $varCache[$index];
2322                         }
2323                 }
2324
2325                 $ts = time();
2326                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2327
2328                 # Use the time zone
2329                 global $wgLocaltimezone;
2330                 if ( isset( $wgLocaltimezone ) ) {
2331                         $oldtz = getenv( 'TZ' );
2332                         putenv( 'TZ='.$wgLocaltimezone );
2333                 }
2334
2335                 wfSuppressWarnings(); // E_STRICT system time bitching
2336                 $localTimestamp = date( 'YmdHis', $ts );
2337                 $localMonth = date( 'm', $ts );
2338                 $localMonthName = date( 'n', $ts );
2339                 $localDay = date( 'j', $ts );
2340                 $localDay2 = date( 'd', $ts );
2341                 $localDayOfWeek = date( 'w', $ts );
2342                 $localWeek = date( 'W', $ts );
2343                 $localYear = date( 'Y', $ts );
2344                 $localHour = date( 'H', $ts );
2345                 if ( isset( $wgLocaltimezone ) ) {
2346                         putenv( 'TZ='.$oldtz );
2347                 }
2348                 wfRestoreWarnings();
2349
2350                 switch ( $index ) {
2351                         case 'currentmonth':
2352                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2353                         case 'currentmonthname':
2354                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2355                         case 'currentmonthnamegen':
2356                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2357                         case 'currentmonthabbrev':
2358                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2359                         case 'currentday':
2360                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2361                         case 'currentday2':
2362                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2363                         case 'localmonth':
2364                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2365                         case 'localmonthname':
2366                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2367                         case 'localmonthnamegen':
2368                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2369                         case 'localmonthabbrev':
2370                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2371                         case 'localday':
2372                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2373                         case 'localday2':
2374                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2375                         case 'pagename':
2376                                 return wfEscapeWikiText( $this->mTitle->getText() );
2377                         case 'pagenamee':
2378                                 return $this->mTitle->getPartialURL();
2379                         case 'fullpagename':
2380                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2381                         case 'fullpagenamee':
2382                                 return $this->mTitle->getPrefixedURL();
2383                         case 'subpagename':
2384                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2385                         case 'subpagenamee':
2386                                 return $this->mTitle->getSubpageUrlForm();
2387                         case 'basepagename':
2388                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2389                         case 'basepagenamee':
2390                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2391                         case 'talkpagename':
2392                                 if( $this->mTitle->canTalk() ) {
2393                                         $talkPage = $this->mTitle->getTalkPage();
2394                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2395                                 } else {
2396                                         return '';
2397                                 }
2398                         case 'talkpagenamee':
2399                                 if( $this->mTitle->canTalk() ) {
2400                                         $talkPage = $this->mTitle->getTalkPage();
2401                                         return $talkPage->getPrefixedUrl();
2402                                 } else {
2403                                         return '';
2404                                 }
2405                         case 'subjectpagename':
2406                                 $subjPage = $this->mTitle->getSubjectPage();
2407                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2408                         case 'subjectpagenamee':
2409                                 $subjPage = $this->mTitle->getSubjectPage();
2410                                 return $subjPage->getPrefixedUrl();
2411                         case 'revisionid':
2412                                 // Let the edit saving system know we should parse the page
2413                                 // *after* a revision ID has been assigned.
2414                                 $this->mOutput->setFlag( 'vary-revision' );
2415                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2416                                 return $this->mRevisionId;
2417                         case 'revisionday':
2418                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2419                         case 'revisionday2':
2420                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2421                         case 'revisionmonth':
2422                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2423                         case 'revisionyear':
2424                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2425                         case 'revisiontimestamp':
2426                                 return $this->getRevisionTimestamp();
2427                         case 'namespace':
2428                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2429                         case 'namespacee':
2430                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2431                         case 'talkspace':
2432                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2433                         case 'talkspacee':
2434                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2435                         case 'subjectspace':
2436                                 return $this->mTitle->getSubjectNsText();
2437                         case 'subjectspacee':
2438                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2439                         case 'currentdayname':
2440                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2441                         case 'currentyear':
2442                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2443                         case 'currenttime':
2444                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2445                         case 'currenthour':
2446                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2447                         case 'currentweek':
2448                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2449                                 // int to remove the padding
2450                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2451                         case 'currentdow':
2452                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2453                         case 'localdayname':
2454                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2455                         case 'localyear':
2456                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2457                         case 'localtime':
2458                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2459                         case 'localhour':
2460                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2461                         case 'localweek':
2462                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2463                                 // int to remove the padding
2464                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2465                         case 'localdow':
2466                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2467                         case 'numberofarticles':
2468                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2469                         case 'numberoffiles':
2470                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2471                         case 'numberofusers':
2472                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2473                         case 'numberofpages':
2474                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2475                         case 'numberofadmins':
2476                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2477                         case 'numberofedits':
2478                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2479                         case 'currenttimestamp':
2480                                 return $varCache[$index] = wfTimestampNow();
2481                         case 'localtimestamp':
2482                                 return $varCache[$index] = $localTimestamp;
2483                         case 'currentversion':
2484                                 return $varCache[$index] = SpecialVersion::getVersion();
2485                         case 'sitename':
2486                                 return $wgSitename;
2487                         case 'server':
2488                                 return $wgServer;
2489                         case 'servername':
2490                                 return $wgServerName;
2491                         case 'scriptpath':
2492                                 return $wgScriptPath;
2493                         case 'directionmark':
2494                                 return $wgContLang->getDirMark();
2495                         case 'contentlanguage':
2496                                 global $wgContLanguageCode;
2497                                 return $wgContLanguageCode;
2498                         default:
2499                                 $ret = null;
2500                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2501                                         return $ret;
2502                                 else
2503                                         return null;
2504                 }
2505         }
2506
2507         /**
2508          * initialise the magic variables (like CURRENTMONTHNAME)
2509          *
2510          * @private
2511          */
2512         function initialiseVariables() {
2513                 $fname = 'Parser::initialiseVariables';
2514                 wfProfileIn( $fname );
2515                 $variableIDs = MagicWord::getVariableIDs();
2516
2517                 $this->mVariables = new MagicWordArray( $variableIDs );
2518                 wfProfileOut( $fname );
2519         }
2520
2521         /**
2522          * Parse any parentheses in format ((title|part|part)} and return the document tree
2523          * This is the ghost of replace_variables().
2524          *
2525          * @param string $text The text to parse
2526          * @private
2527          */
2528         function preprocessToDom ( $text ) {
2529                 wfProfileIn( __METHOD__ );
2530                 wfProfileIn( __METHOD__.'-makexml' );
2531
2532                 static $msgRules, $normalRules;
2533                 if ( !$msgRules ) {
2534                         $msgRules = array(
2535                                 '{' => array(
2536                                         'end' => '}',
2537                                         'names' => array(
2538                                                 2 => 'template',
2539                                         ),
2540                                         'min' => 2,
2541                                         'max' => 2,
2542                                 ),
2543                                 '[' => array(
2544                                         'end' => ']',
2545                                         'names' => array( 2 => null ),
2546                                         'min' => 2,
2547                                         'max' => 2,
2548                                 )
2549                         );
2550                         $normalRules = array(
2551                                 '{' => array(
2552                                         'end' => '}',
2553                                         'names' => array(
2554                                                 2 => 'template',
2555                                                 3 => 'tplarg',
2556                                         ),
2557                                         'min' => 2,
2558                                         'max' => 3,
2559                                 ),
2560                                 '[' => array(
2561                                         'end' => ']',
2562                                         'names' => array( 2 => null ),
2563                                         'min' => 2,
2564                                         'max' => 2,
2565                                 )
2566                         );
2567                 }
2568                 if ( $this->ot['msg'] ) {
2569                         $rules = $msgRules;
2570                 } else {
2571                         $rules = $normalRules;
2572                 }
2573
2574                 $extElements = implode( '|', $this->getStripList() );
2575                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2576                 $extElementsRegex = "/($extElements)(?:\s|\/>|>)|(!--)/iA";
2577
2578                 $stack = array();      # Stack of unclosed parentheses
2579                 $stackIndex = -1;      # Stack read pointer
2580
2581                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2582
2583                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2584                 $topAccum = '<root>';      # Top level text accumulator
2585                 $accum =& $topAccum;            # Current text accumulator
2586                 $findEquals = false;            # True to find equals signs in arguments
2587                 $findHeading = false;           # True to look at LF characters for possible headings
2588                 $findPipe = false;              # True to take notice of pipe characters
2589                 $headingIndex = 1;
2590                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2591
2592                 while ( $i < strlen( $text ) ) {
2593                         if ( $i == -1 ) {
2594                                 $found = 'line-start';
2595                                 $curChar = '';
2596                         } else {
2597                                 # Find next opening brace, closing brace or pipe
2598                                 $search = $searchBase;
2599                                 if ( $stackIndex == -1 ) {
2600                                         $currentClosing = '';
2601                                         // Look for headings only at the top stack level
2602                                         // Among other things, this resolves the ambiguity between =
2603                                         // for headings and = for template arguments
2604                                         $search .= "\n";
2605                                 } else {
2606                                         $currentClosing = $stack[$stackIndex]['close'];
2607                                         $search .= $currentClosing;
2608                                 }
2609                                 if ( $findPipe ) {
2610                                         $search .= '|';
2611                                 }
2612                                 if ( $findEquals ) {
2613                                         $search .= '=';
2614                                 }
2615                                 $rule = null;
2616                                 # Output literal section, advance input counter
2617                                 $literalLength = strcspn( $text, $search, $i );
2618                                 if ( $literalLength > 0 ) {
2619                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2620                                         $i += $literalLength;
2621                                 }
2622                                 if ( $i >= strlen( $text ) ) {
2623                                         if ( $currentClosing == "\n" ) {
2624                                                 // Do a past-the-end run to finish off the heading
2625                                                 $curChar = '';
2626                                                 $found = 'line-end';
2627                                         } else {
2628                                                 # All done
2629                                                 break;
2630                                         }
2631                                 } else {
2632                                         $curChar = $text[$i];
2633                                         if ( $curChar == '|' ) {
2634                                                 $found = 'pipe';
2635                                         } elseif ( $curChar == '=' ) {
2636                                                 $found = 'equals';
2637                                         } elseif ( $curChar == '<' ) {
2638                                                 $found = 'angle';
2639                                         } elseif ( $curChar == "\n" ) {
2640                                                 if ( $stackIndex == -1 ) {
2641                                                         $found = 'line-start';
2642                                                 } else {
2643                                                         $found = 'line-end';
2644                                                 }
2645                                         } elseif ( $curChar == $currentClosing ) {
2646                                                 $found = 'close';
2647                                         } elseif ( isset( $rules[$curChar] ) ) {
2648                                                 $found = 'open';
2649                                                 $rule = $rules[$curChar];
2650                                         } else {
2651                                                 # Some versions of PHP have a strcspn which stops on null characters
2652                                                 # Ignore and continue
2653                                                 ++$i;
2654                                                 continue;
2655                                         }
2656                                 }
2657                         }
2658
2659                         if ( $found == 'angle' ) {
2660                                 $matches = false;
2661                                 // Determine element name
2662                                 if ( !preg_match( $extElementsRegex, $text, $matches, 0, $i + 1 ) ) {
2663                                         // Element name missing or not listed
2664                                         $accum .= '&lt;';
2665                                         ++$i;
2666                                         continue;
2667                                 }
2668                                 // Handle comments
2669                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2670                                         // HTML comment, scan to end
2671                                         $endpos = strpos( $text, '-->', $i + 4 );
2672                                         if ( $endpos === false ) {
2673                                                 // Unclosed comment in input, runs to end
2674                                                 $accum .= htmlspecialchars( substr( $text, $i ) );
2675                                                 if ( $this->ot['html'] ) {
2676                                                         // Close it so later stripping can remove it
2677                                                         $accum .= htmlspecialchars( '-->' );
2678                                                 }
2679                                                 $i = strlen( $text );
2680                                                 continue;
2681                                         }
2682                                         $accum .= htmlspecialchars( substr( $text, $i, $endpos - $i + 3 ) );
2683                                         #$inner = substr( $text, $i + 4, $endpos - $i - 4 );
2684                                         #$accum .= '<ext><name>!--</name><inner>' . htmlspecialchars( $inner ) . '</inner></ext>';
2685                                         $i = $endpos + 3;
2686                                         continue;
2687                                 }
2688                                 $name = $matches[1];
2689                                 $attrStart = $i + strlen( $name ) + 1;
2690
2691                                 // Find end of tag
2692                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2693                                 if ( $tagEndPos === false ) {
2694                                         // Infinite backtrack
2695                                         // Disable tag search to prevent worst-case O(N^2) performance
2696                                         $noMoreGT = true;
2697                                         $accum .= '&lt;';
2698                                         ++$i;
2699                                         continue;
2700                                 }
2701                                 if ( $text[$tagEndPos-1] == '/' ) {
2702                                         $attrEnd = $tagEndPos - 1;
2703                                         $inner = null;
2704                                         $i = $tagEndPos + 1;
2705                                         $close = '';
2706                                 } else {
2707                                         $attrEnd = $tagEndPos;
2708                                         // Find closing tag
2709                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2710                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2711                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2712                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2713                                         } else {
2714                                                 // No end tag -- let it run out to the end of the text.
2715                                                 $inner = substr( $text, $tagEndPos + 1 );
2716                                                 $i = strlen( $text );
2717                                                 $close = '';
2718                                         }
2719                                 }
2720                                 $accum .= '<ext>';
2721                                 if ( $attrEnd <= $attrStart ) {
2722                                         $attr = '';
2723                                 } else {
2724                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2725                                 }
2726                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2727                                         // Note that the attr element contains the whitespace between name and attribute,
2728                                         // this is necessary for precise reconstruction during pre-save transform.
2729                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2730                                 if ( $inner !== null ) {
2731                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2732                                 }
2733                                 $accum .= $close . '</ext>';
2734                         }
2735
2736                         elseif ( $found == 'line-start' ) {
2737                                 // Is this the start of a heading?
2738                                 // Line break belongs before the heading element in any case
2739                                 $accum .= $curChar;
2740                                 $i++;
2741
2742                                 $count = strspn( $text, '=', $i, 6 );
2743                                 if ( $count > 0 ) {
2744                                         $piece = array(
2745                                                 'open' => "\n",
2746                                                 'close' => "\n",
2747                                                 'parts' => array( str_repeat( '=', $count ) ),
2748                                                 'count' => $count );
2749                                         $stack[++$stackIndex] = $piece;
2750                                         $i += $count;
2751                                         $accum =& $stack[$stackIndex]['parts'][0];
2752                                         $findPipe = false;
2753                                 }
2754                         }
2755
2756                         elseif ( $found == 'line-end' ) {
2757                                 $piece = $stack[$stackIndex];
2758                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2759                                 assert( $piece['open'] == "\n" );
2760                                 assert( $stackIndex == 0 );
2761                                 // Search back through the accumulator to see if it has a proper close
2762                                 // No efficient way to do this in PHP AFAICT: strrev, PCRE search with $ anchor
2763                                 // and rtrim are all O(N) in total size. Optimal would be O(N) in trailing
2764                                 // whitespace size only.
2765                                 $m = false;
2766                                 $count = $piece['count'];
2767                                 if ( preg_match( "/(={{$count}})\s*$/", $accum, $m, 0, $count ) ) {
2768                                         // Found match, output <h>
2769                                         $count = min( strlen( $m[1] ), $count );
2770                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2771                                         $headingIndex++;
2772                                 } else {
2773                                         // No match, no <h>, just pass down the inner text
2774                                         $element = $accum;
2775                                 }
2776                                 // Unwind the stack
2777                                 // Headings can only occur on the top level, so this is a bit simpler than the
2778                                 // generic stack unwind operation in the close case
2779                                 unset( $stack[$stackIndex--] );
2780                                 $accum =& $topAccum;
2781                                 $findEquals = false;
2782                                 $findPipe = false;
2783
2784                                 // Append the result to the enclosing accumulator
2785                                 $accum .= $element;
2786                                 // Note that we do NOT increment the input pointer.
2787                                 // This is because the closing linebreak could be the opening linebreak of
2788                                 // another heading. Infinite loops are avoided because the next iteration MUST
2789                                 // hit the heading open case above, which unconditionally increments the
2790                                 // input pointer.
2791                         }
2792
2793                         elseif ( $found == 'open' ) {
2794                                 # count opening brace characters
2795                                 $count = strspn( $text, $curChar, $i );
2796
2797                                 # we need to add to stack only if opening brace count is enough for one of the rules
2798                                 if ( $count >= $rule['min'] ) {
2799                                         # Add it to the stack
2800                                         $piece = array(
2801                                                 'open' => $curChar,
2802                                                 'close' => $rule['end'],
2803                                                 'count' => $count,
2804                                                 'parts' => array( '' ),
2805                                                 'eqpos' => array(),
2806                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2807                                         );
2808
2809                                         $stackIndex ++;
2810                                         $stack[$stackIndex] = $piece;
2811                                         $accum =& $stack[$stackIndex]['parts'][0];
2812                                         $findEquals = false;
2813                                         $findPipe = true;
2814                                 } else {
2815                                         # Add literal brace(s)
2816                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2817                                 }
2818                                 $i += $count;
2819                         }
2820
2821                         elseif ( $found == 'close' ) {
2822                                 $piece = $stack[$stackIndex];
2823                                 # lets check if there are enough characters for closing brace
2824                                 $maxCount = $piece['count'];
2825                                 $count = strspn( $text, $curChar, $i, $maxCount );
2826
2827                                 # check for maximum matching characters (if there are 5 closing
2828                                 # characters, we will probably need only 3 - depending on the rules)
2829                                 $matchingCount = 0;
2830                                 $rule = $rules[$piece['open']];
2831                                 if ( $count > $rule['max'] ) {
2832                                         # The specified maximum exists in the callback array, unless the caller
2833                                         # has made an error
2834                                         $matchingCount = $rule['max'];
2835                                 } else {
2836                                         # Count is less than the maximum
2837                                         # Skip any gaps in the callback array to find the true largest match
2838                                         # Need to use array_key_exists not isset because the callback can be null
2839                                         $matchingCount = $count;
2840                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2841                                                 --$matchingCount;
2842                                         }
2843                                 }
2844
2845                                 if ($matchingCount <= 0) {
2846                                         # No matching element found in callback array
2847                                         # Output a literal closing brace and continue
2848                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2849                                         $i += $count;
2850                                         continue;
2851                                 }
2852                                 $name = $rule['names'][$matchingCount];
2853                                 if ( $name === null ) {
2854                                         // No element, just literal text
2855                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2856                                                 implode( '|', $piece['parts'] ) .
2857                                                 str_repeat( $rule['end'], $matchingCount );
2858                                 } else {
2859                                         # Create XML element
2860                                         # Note: $parts is already XML, does not need to be encoded further
2861                                         $parts = $piece['parts'];
2862                                         $title = $parts[0];
2863                                         unset( $parts[0] );
2864
2865                                         # The invocation is at the start of the line if lineStart is set in
2866                                         # the stack, and all opening brackets are used up.
2867                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
2868                                                 $attr = ' lineStart="1"';
2869                                         } else {
2870                                                 $attr = '';
2871                                         }
2872
2873                                         $element = "<$name$attr>";
2874                                         $element .= "<title>$title</title>";
2875                                         $argIndex = 1;
2876                                         foreach ( $parts as $partIndex => $part ) {
2877                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
2878                                                         $eqpos = $piece['eqpos'][$partIndex];
2879                                                         list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
2880                                                         list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
2881                                                         $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
2882                                                 } else {
2883                                                         list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
2884                                                         $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
2885                                                         $argIndex++;
2886                                                 }
2887                                         }
2888                                         $element .= "</$name>";
2889                                 }
2890
2891                                 # Advance input pointer
2892                                 $i += $matchingCount;
2893
2894                                 # Unwind the stack
2895                                 unset( $stack[$stackIndex--] );
2896                                 if ( $stackIndex == -1 ) {
2897                                         $accum =& $topAccum;
2898                                         $findEquals = false;
2899                                         $findPipe = false;
2900                                 } else {
2901                                         $partCount = count( $stack[$stackIndex]['parts'] );
2902                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
2903                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
2904                                         $findEquals = $findPipe && $partCount > 1
2905                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
2906                                 }
2907
2908                                 # Re-add the old stack element if it still has unmatched opening characters remaining
2909                                 if ($matchingCount < $piece['count']) {
2910                                         $piece['parts'] = array( '' );
2911                                         $piece['count'] -= $matchingCount;
2912                                         $piece['eqpos'] = array();
2913                                         # do we still qualify for any callback with remaining count?
2914                                         $names = $rules[$piece['open']]['names'];
2915                                         $skippedBraces = 0;
2916                                         $enclosingAccum =& $accum;
2917                                         while ( $piece['count'] ) {
2918                                                 if ( array_key_exists( $piece['count'], $names ) ) {
2919                                                         $stackIndex++;
2920                                                         $stack[$stackIndex] = $piece;
2921                                                         $accum =& $stack[$stackIndex]['parts'][0];
2922                                                         $findEquals = true;
2923                                                         $findPipe = true;
2924                                                         break;
2925                                                 }
2926                                                 --$piece['count'];
2927                                                 $skippedBraces ++;
2928                                         }
2929                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
2930                                 }
2931
2932                                 # Add XML element to the enclosing accumulator
2933                                 $accum .= $element;
2934                         }
2935
2936                         elseif ( $found == 'pipe' ) {
2937                                 $stack[$stackIndex]['parts'][] = '';
2938                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2939                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
2940                                 $findEquals = true;
2941                                 ++$i;
2942                         }
2943
2944                         elseif ( $found == 'equals' ) {
2945                                 $findEquals = false;
2946                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2947                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
2948                                 $accum .= '=';
2949                                 ++$i;
2950                         }
2951                 }
2952
2953                 # Output any remaining unclosed brackets
2954                 foreach ( $stack as $piece ) {
2955                         if ( $piece['open'] == "\n" ) {
2956                                 $topAccum .= $piece['parts'][0];
2957                         } else {
2958                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
2959                         }
2960                 }
2961                 $topAccum .= '</root>';
2962
2963                 wfProfileOut( __METHOD__.'-makexml' );
2964                 wfProfileIn( __METHOD__.'-loadXML' );
2965                 $dom = new DOMDocument;
2966                 if ( !$dom->loadXML( $topAccum ) ) {
2967                         throw new MWException( __METHOD__.' generated invalid XML' );
2968                 }
2969                 wfProfileOut( __METHOD__.'-loadXML' );
2970                 wfProfileOut( __METHOD__ );
2971                 return $dom;
2972         }
2973
2974         /*
2975          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2976          */
2977         public static function splitWhitespace( $s ) {
2978                 $ltrimmed = ltrim( $s );
2979                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2980                 $trimmed = rtrim( $ltrimmed );
2981                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2982                 if ( $diff > 0 ) {
2983                         $w2 = substr( $ltrimmed, -$diff );
2984                 } else {
2985                         $w2 = '';
2986                 }
2987                 return array( $w1, $trimmed, $w2 );
2988         }
2989
2990         /**
2991          * Replace magic variables, templates, and template arguments
2992          * with the appropriate text. Templates are substituted recursively,
2993          * taking care to avoid infinite loops.
2994          *
2995          * Note that the substitution depends on value of $mOutputType:
2996          *  OT_WIKI: only {{subst:}} templates
2997          *  OT_MSG: only magic variables
2998          *  OT_HTML: all templates and magic variables
2999          *
3000          * @param string $tex The text to transform
3001          * @param PPFrame $frame Object describing the arguments passed to the template
3002          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3003          * @private
3004          */
3005         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3006                 # Prevent too big inclusions
3007                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3008                         return $text;
3009                 }
3010
3011                 $fname = __METHOD__;
3012                 wfProfileIn( $fname );
3013
3014                 if ( $frame === false ) {
3015                         $frame = new PPFrame( $this );
3016                 } elseif ( !( $frame instanceof PPFrame ) ) {
3017                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3018                 }
3019
3020                 $dom = $this->preprocessToDom( $text );
3021                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3022                 $text = $frame->expand( $dom, 0, $flags );
3023
3024                 wfProfileOut( $fname );
3025                 return $text;
3026         }
3027
3028         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3029         static function createAssocArgs( $args ) {
3030                 $assocArgs = array();
3031                 $index = 1;
3032                 foreach( $args as $arg ) {
3033                         $eqpos = strpos( $arg, '=' );
3034                         if ( $eqpos === false ) {
3035                                 $assocArgs[$index++] = $arg;
3036                         } else {
3037                                 $name = trim( substr( $arg, 0, $eqpos ) );
3038                                 $value = trim( substr( $arg, $eqpos+1 ) );
3039                                 if ( $value === false ) {
3040                                         $value = '';
3041                                 }
3042                                 if ( $name !== false ) {
3043                                         $assocArgs[$name] = $value;
3044                                 }
3045                         }
3046                 }
3047
3048                 return $assocArgs;
3049         }
3050
3051         /**
3052          * Return the text of a template, after recursively
3053          * replacing any variables or templates within the template.
3054          *
3055          * @param array $piece The parts of the template
3056          *  $piece['text']: matched text
3057          *  $piece['title']: the title, i.e. the part before the |
3058          *  $piece['parts']: the parameter array
3059          * @param PPFrame The current frame, contains template arguments
3060          * @return string the text of the template
3061          * @private
3062          */
3063         function braceSubstitution( $piece, $frame ) {
3064                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3065                 $fname = __METHOD__;
3066                 wfProfileIn( $fname );
3067                 wfProfileIn( __METHOD__.'-setup' );
3068
3069                 # Flags
3070                 $found = false;             # $text has been filled
3071                 $nowiki = false;            # wiki markup in $text should be escaped
3072                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
3073                 $noargs = false;            # Don't replace triple-brace arguments in $text
3074                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3075                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3076                 $isDOM = false;             # $text is a DOM node needing expansion
3077
3078                 # Title object, where $text came from
3079                 $title = NULL;
3080
3081                 # $part1 is the bit before the first |, and must contain only title characters.
3082                 # Various prefixes will be stripped from it later.
3083                 $titleWithSpaces = $frame->expand( $piece['title'] );
3084                 $part1 = trim( $titleWithSpaces );
3085                 $titleText = false;
3086
3087                 # Original title text preserved for various purposes
3088                 $originalTitle = $part1;
3089
3090                 # $args is a list of argument nodes, starting from index 0, not including $part1
3091                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3092                 wfProfileOut( __METHOD__.'-setup' );
3093
3094                 # SUBST
3095                 wfProfileIn( __METHOD__.'-modifiers' );
3096                 if ( !$found ) {
3097                         $mwSubst =& MagicWord::get( 'subst' );
3098                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3099                                 # One of two possibilities is true:
3100                                 # 1) Found SUBST but not in the PST phase
3101                                 # 2) Didn't find SUBST and in the PST phase
3102                                 # In either case, return without further processing
3103                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3104                                 $found = true;
3105                                 $noparse = true;
3106                                 $noargs = true;
3107                         }
3108                 }
3109
3110                 # Variables
3111                 if ( !$found && $args->length == 0 ) {
3112                         $id = $this->mVariables->matchStartToEnd( $part1 );
3113                         if ( $id !== false ) {
3114                                 $text = $this->getVariableValue( $id );
3115                                 $this->mOutput->mContainsOldMagic = true;
3116                                 $found = true;
3117                                 $noparse = true;
3118                                 $noargs = true;
3119                         }
3120                 }
3121
3122                 # MSG, MSGNW and RAW
3123                 if ( !$found ) {
3124                         # Check for MSGNW:
3125                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3126                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3127                                 $nowiki = true;
3128                         } else {
3129                                 # Remove obsolete MSG:
3130                                 $mwMsg =& MagicWord::get( 'msg' );
3131                                 $mwMsg->matchStartAndRemove( $part1 );
3132                         }
3133
3134                         # Check for RAW:
3135                         $mwRaw =& MagicWord::get( 'raw' );
3136                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3137                                 $forceRawInterwiki = true;
3138                         }
3139                 }
3140                 wfProfileOut( __METHOD__.'-modifiers' );
3141
3142                 # Save path level before recursing into functions & templates.
3143                 $lastPathLevel = $this->mTemplatePath;
3144
3145                 # Parser functions
3146                 if ( !$found ) {
3147                         wfProfileIn( __METHOD__ . '-pfunc' );
3148
3149                         $colonPos = strpos( $part1, ':' );
3150                         if ( $colonPos !== false ) {
3151                                 # Case sensitive functions
3152                                 $function = substr( $part1, 0, $colonPos );
3153                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3154                                         $function = $this->mFunctionSynonyms[1][$function];
3155                                 } else {
3156                                         # Case insensitive functions
3157                                         $function = strtolower( $function );
3158                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3159                                                 $function = $this->mFunctionSynonyms[0][$function];
3160                                         } else {
3161                                                 $function = false;
3162                                         }
3163                                 }
3164                                 if ( $function ) {
3165                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3166                                         $initialArgs = array( &$this );
3167                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3168                                         if ( $flags & SFH_OBJECT_ARGS ) {
3169                                                 # Add a frame parameter, and pass the arguments as an array
3170                                                 $allArgs = $initialArgs;
3171                                                 $allArgs[] = $frame;
3172                                                 foreach ( $args as $arg ) {
3173                                                         $funcArgs[] = $arg;
3174                                                 }
3175                                                 $allArgs[] = $funcArgs;
3176                                         } else {
3177                                                 # Convert arguments to plain text
3178                                                 foreach ( $args as $arg ) {
3179                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3180                                                 }
3181                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3182                                         }
3183
3184                                         if (! is_callable($callback)) {
3185                                                 if (is_array($callback))
3186                                                         $callback = $callback[0];
3187                                                 else
3188                                                         die ("\nInvalid callback for $function: '$callback' ($flags)\n");
3189                                         }
3190
3191                                         $result = call_user_func_array( $callback, $allArgs );
3192                                         $found = true;
3193
3194                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3195                                         $noargs = true;
3196                                         $noparse = true;
3197
3198                                         if ( is_array( $result ) ) {
3199                                                 if ( isset( $result[0] ) ) {
3200                                                         $text = $result[0];
3201                                                         unset( $result[0] );
3202                                                 }
3203
3204                                                 // Extract flags into the local scope
3205                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3206                                                 extract( $result );
3207                                         } else {
3208                                                 $text = $result;
3209                                         }
3210                                 }
3211                         }
3212                         wfProfileOut( __METHOD__ . '-pfunc' );
3213                 }
3214
3215                 # Finish mangling title and then check for loops.
3216                 # Set $title to a Title object and $titleText to the PDBK
3217                 if ( !$found ) {
3218                         $ns = NS_TEMPLATE;
3219                         # Split the title into page and subpage
3220                         $subpage = '';
3221                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3222                         if ($subpage !== '') {
3223                                 $ns = $this->mTitle->getNamespace();
3224                         }
3225                         $title = Title::newFromText( $part1, $ns );
3226                         if ( $title ) {
3227                                 $titleText = $title->getPrefixedText();
3228                                 # Check for language variants if the template is not found
3229                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3230                                         $wgContLang->findVariantLink($part1, $title);
3231                                 }
3232                                 # Do infinite loop check
3233                                 if ( isset( $this->mTemplatePath[$titleText] ) ) {
3234                                         $noparse = true;
3235                                         $noargs = true;
3236                                         $found = true;
3237                                         $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
3238                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3239                                 }
3240                         }
3241                 }
3242
3243                 # Load from database
3244                 if ( !$found && $title ) {
3245                         wfProfileIn( __METHOD__ . '-loadtpl' );
3246                         if ( !$title->isExternal() ) {
3247                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3248                                         $text = SpecialPage::capturePath( $title );
3249                                         if ( is_string( $text ) ) {
3250                                                 $found = true;
3251                                                 $noparse = true;
3252                                                 $noargs = true;
3253                                                 $isHTML = true;
3254                                                 $this->disableCache();
3255                                         }
3256                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3257                                         $found = false; //access denied
3258                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3259                                 } else {
3260                                         list( $text, $title ) = $this->getTemplateDom( $title );
3261                                         if ( $text !== false ) {
3262                                                 $found = true;
3263                                                 $isDOM = true;
3264                                         }
3265                                 }
3266
3267                                 # If the title is valid but undisplayable, make a link to it
3268                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3269                                         $text = "[[:$titleText]]";
3270                                         $found = true;
3271                                 }
3272                         } elseif ( $title->isTrans() ) {
3273                                 // Interwiki transclusion
3274                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3275                                         $text = $this->interwikiTransclude( $title, 'render' );
3276                                         $isHTML = true;
3277                                         $noparse = true;
3278                                 } else {
3279                                         $text = $this->interwikiTransclude( $title, 'raw' );
3280                                 }
3281                                 $found = true;
3282                         }
3283                         wfProfileOut( __METHOD__ . '-loadtpl' );
3284                 }
3285
3286                 # Recursive parsing, escaping and link table handling
3287                 # Only for HTML output
3288                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3289                         if ( $isDOM ) {
3290                                 $text = $frame->expand( $text );
3291                         }
3292                         $text = wfEscapeWikiText( $text );
3293                 } elseif ( !$this->ot['msg'] && $found ) {
3294                         if ( $noargs ) {
3295                                 $newFrame = $frame->newChild();
3296                         } else {
3297                                 # Clean up argument array
3298                                 $newFrame = $frame->newChild( $args, $title );
3299                                 # Add a new element to the templace recursion path
3300                                 $this->mTemplatePath[$titleText] = 1;
3301                         }
3302
3303                         if ( !$noparse ) {
3304                                 if ( $isDOM ) {
3305                                         if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
3306                                                 # Expansion is eligible for the empty-frame cache
3307                                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3308                                                         $text = $this->mTplExpandCache[$titleText];
3309                                                 } else {
3310                                                         $text = $newFrame->expand( $text );
3311                                                         $this->mTplExpandCache[$titleText] = $text;
3312                                                 }
3313                                         } else {
3314                                                 $text = $newFrame->expand( $text );
3315                                         }
3316                                 } else {
3317                                         $text = $this->replaceVariables( $text, $newFrame );
3318                                 }
3319
3320                                 # strip woz 'ere 2004-07
3321
3322                                 # Bug 529: if the template begins with a table or block-level
3323                                 # element, it should be treated as beginning a new line.
3324                                 # This behaviour is somewhat controversial.
3325                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3326                                         $text = "\n" . $text;
3327                                 }
3328                         } elseif ( !$noargs ) {
3329                                 # $noparse and !$noargs
3330                                 # Just replace the arguments, not any double-brace items
3331                                 # This is used for rendered interwiki transclusion
3332                                 if ( $isDOM ) {
3333                                         $text = $newFrame->expand( $text, 0, PPFrame::NO_TEMPLATES );
3334                                 } else {
3335                                         $text = $this->replaceVariables( $text, $newFrame, true );
3336                                 }
3337                         } elseif ( $isDOM ) {
3338                                 $text = $frame->expand( $text );
3339                         }
3340                 } elseif ( $isDOM ) {
3341                         $text = $frame->expand( $text, 0, PPFrame::NO_TEMPLATES | PPFrame::NO_ARGS );
3342                 }
3343
3344                 # Prune lower levels off the recursion check path
3345                 $this->mTemplatePath = $lastPathLevel;
3346
3347                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3348                         # Error, oversize inclusion
3349                         $text = "[[$originalTitle]]" .
3350                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3351                         $noparse = true;
3352                         $noargs = true;
3353                 }
3354
3355                 if ( !$found ) {
3356                         wfProfileOut( $fname );
3357                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3358                 } else {
3359                         wfProfileIn( __METHOD__ . '-placeholders' );
3360                         if ( $isHTML ) {
3361                                 # Replace raw HTML by a placeholder
3362                                 # Add a blank line preceding, to prevent it from mucking up
3363                                 # immediately preceding headings
3364                                 $text = "\n\n" . $this->insertStripItem( $text );
3365                         }
3366                         wfProfileOut( __METHOD__ . '-placeholders' );
3367                 }
3368
3369                 # Prune lower levels off the recursion check path
3370                 $this->mTemplatePath = $lastPathLevel;
3371
3372                 if ( !$found ) {
3373                         wfProfileOut( $fname );
3374                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3375                 } else {
3376                         wfProfileOut( $fname );
3377                         return $text;
3378                 }
3379         }
3380
3381         /**
3382          * Get the semi-parsed DOM representation of a template with a given title,
3383          * and its redirect destination title. Cached.
3384          */
3385         function getTemplateDom( $title ) {
3386                 $titleText = $title->getPrefixedDBkey();
3387
3388                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3389                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3390                         $title = Title::makeTitle( $ns, $dbk );
3391                         $titleText = $title->getPrefixedDBkey();
3392                 }
3393                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3394                         return array( $this->mTplDomCache[$titleText], $title );
3395                 }
3396
3397                 // Cache miss, go to the database
3398                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3399
3400                 if ( $text === false ) {
3401                         $this->mTplDomCache[$titleText] = false;
3402                         return array( false, $title );
3403                 }
3404
3405                 # If there are any <onlyinclude> tags, only include them
3406                 if ( !$this->ot['msg'] ) {
3407                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3408                                 $replacer = new OnlyIncludeReplacer;
3409                                 StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3410                                         array( &$replacer, 'replace' ), $text );
3411                                 $text = $replacer->output;
3412                         }
3413                         # Remove <noinclude> sections and <includeonly> tags
3414                         $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3415                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3416                 }
3417
3418                 $dom = $this->preprocessToDom( $text );
3419                 $this->mTplDomCache[$titleText] = $dom;
3420                 return array( $dom, $title );
3421         }
3422
3423         /**
3424          * Fetch the unparsed text of a template and register a reference to it.
3425          */
3426         function fetchTemplateAndTitle( $title ) {
3427                 $templateCb = $this->mOptions->getTemplateCallback();
3428                 $stuff = call_user_func( $templateCb, $title );
3429                 $text = $stuff['text'];
3430                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3431                 if ( isset( $stuff['deps'] ) ) {
3432                         foreach ( $stuff['deps'] as $dep ) {
3433                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3434                         }
3435                 }
3436                 return array($text,$finalTitle);
3437         }
3438
3439         function fetchTemplate( $title ) {
3440                 $rv = $this->fetchTemplateAndTitle($title);
3441                 return $rv[0];
3442         }
3443
3444         /**
3445          * Static function to get a template
3446          * Can be overridden via ParserOptions::setTemplateCallback().
3447          */
3448         static function statelessFetchTemplate( $title ) {
3449                 $text = $skip = false;
3450                 $finalTitle = $title;
3451                 $deps = array();
3452
3453                 // Loop to fetch the article, with up to 1 redirect
3454                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3455                         # Give extensions a chance to select the revision instead
3456                         $id = false; // Assume current
3457                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3458
3459                         if( $skip ) {
3460                                 $text = false;
3461                                 $deps[] = array(
3462                                         'title' => $title,
3463                                         'page_id' => $title->getArticleID(),
3464                                         'rev_id' => null );
3465                                 break;
3466                         }
3467                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3468                         $rev_id = $rev ? $rev->getId() : 0;
3469
3470                         $deps[] = array(
3471                                 'title' => $title,
3472                                 'page_id' => $title->getArticleID(),
3473                                 'rev_id' => $rev_id );
3474
3475                         if( $rev ) {
3476                                 $text = $rev->getText();
3477                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3478                                 global $wgLang;
3479                                 $message = $wgLang->lcfirst( $title->getText() );
3480                                 $text = wfMsgForContentNoTrans( $message );
3481                                 if( wfEmptyMsg( $message, $text ) ) {
3482                                         $text = false;
3483                                         break;
3484                                 }
3485                         } else {
3486                                 break;
3487                         }
3488                         if ( $text === false ) {
3489                                 break;
3490                         }
3491                         // Redirect?
3492                         $finalTitle = $title;
3493                         $title = Title::newFromRedirect( $text );
3494                 }
3495                 return array(
3496                         'text' => $text,
3497                         'finalTitle' => $finalTitle,
3498                         'deps' => $deps );
3499         }
3500
3501         /**
3502          * Transclude an interwiki link.
3503          */
3504         function interwikiTransclude( $title, $action ) {
3505                 global $wgEnableScaryTranscluding;
3506
3507                 if (!$wgEnableScaryTranscluding)
3508                         return wfMsg('scarytranscludedisabled');
3509
3510                 $url = $title->getFullUrl( "action=$action" );
3511
3512                 if (strlen($url) > 255)
3513                         return wfMsg('scarytranscludetoolong');
3514                 return $this->fetchScaryTemplateMaybeFromCache($url);
3515         }
3516
3517         function fetchScaryTemplateMaybeFromCache($url) {
3518                 global $wgTranscludeCacheExpiry;
3519                 $dbr = wfGetDB(DB_SLAVE);
3520                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3521                                 array('tc_url' => $url));
3522                 if ($obj) {
3523                         $time = $obj->tc_time;
3524                         $text = $obj->tc_contents;
3525                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3526                                 return $text;
3527                         }
3528                 }
3529
3530                 $text = Http::get($url);
3531                 if (!$text)
3532                         return wfMsg('scarytranscludefailed', $url);
3533
3534                 $dbw = wfGetDB(DB_MASTER);
3535                 $dbw->replace('transcache', array('tc_url'), array(
3536                         'tc_url' => $url,
3537                         'tc_time' => time(),
3538                         'tc_contents' => $text));
3539                 return $text;
3540         }
3541
3542
3543         /**
3544          * Triple brace replacement -- used for template arguments
3545          * @private
3546          */
3547         function argSubstitution( $piece, $frame ) {
3548                 wfProfileIn( __METHOD__ );
3549
3550                 $text = false;
3551                 $error = false;
3552                 $parts = $piece['parts'];
3553                 $argWithSpaces = $frame->expand( $piece['title'] );
3554                 $arg = trim( $argWithSpaces );
3555
3556                 if ( isset( $frame->args[$arg] ) ) {
3557                         $text = $frame->parent->expand( $frame->args[$arg] );
3558                 } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3559                         $text = $frame->expand( $parts->item( 0 ) );
3560                 }
3561                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3562                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3563                 }
3564
3565                 if ( $text === false ) {
3566                         $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
3567                 }
3568                 if ( $error !== false ) {
3569                         $text .= $error;
3570                 }
3571
3572                 wfProfileOut( __METHOD__ );
3573                 return $text;
3574         }
3575
3576         /**
3577          * Return the text to be used for a given extension tag.
3578          * This is the ghost of strip().
3579          *
3580          * @param array $params Associative array of parameters:
3581          *     name       DOMNode for the tag name
3582          *     attrText   DOMNode for unparsed text where tag attributes are thought to be
3583          *     inner      Contents of extension element
3584          *     noClose    Original text did not have a close tag
3585          * @param PPFrame $frame
3586          */
3587         function extensionSubstitution( $params, $frame ) {
3588                 global $wgRawHtml, $wgContLang;
3589                 static $n = 1;
3590
3591                 $name = $frame->expand( $params['name'] );
3592                 $attrText = is_null( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3593                 $content = is_null( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3594
3595                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3596
3597                 if ( $this->ot['html'] ) {
3598                         if ( $name == '!--' ) {
3599                                 return '';
3600                         }
3601                         $name = strtolower( $name );
3602
3603                         $params = Sanitizer::decodeTagAttributes( $attrText );
3604                         switch ( $name ) {
3605                                 case 'html':
3606                                         if( $wgRawHtml ) {
3607                                                 $output = $content;
3608                                                 break;
3609                                         } else {
3610                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3611                                         }
3612                                 case 'nowiki':
3613                                         $output = Xml::escapeTagsOnly( $content );
3614                                         break;
3615                                 case 'math':
3616                                         $output = $wgContLang->armourMath(
3617                                                 MathRenderer::renderMath( $content, $params ) );
3618                                         break;
3619                                 case 'gallery':
3620                                         $output = $this->renderImageGallery( $content, $params );
3621                                         break;
3622                                 default:
3623                                         if( isset( $this->mTagHooks[$name] ) ) {
3624                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3625                                                         array( $content, $params, $this ) );
3626                                         } else {
3627                                                 throw new MWException( "Invalid call hook $name" );
3628                                         }
3629                         }
3630                 } else {
3631                         if ( $name == '!--' ) {
3632                                 $output = '<!--' . $content . '-->';
3633                         } else {
3634                                 if ( $content === null ) {
3635                                         $output = "<$name$attrText/>";
3636                                 } else {
3637                                         $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3638                                         $output = "<$name$attrText>$content$close";
3639                                 }
3640                         }
3641                 }
3642
3643                 if ( $name == 'html' || $name == 'nowiki' ) {
3644                         $this->mStripState->nowiki->setPair( $marker, $output );
3645                 } else {
3646                         $this->mStripState->general->setPair( $marker, $output );
3647                 }
3648                 return $marker;
3649         }
3650
3651         /**
3652          * Increment an include size counter
3653          *
3654          * @param string $type The type of expansion
3655          * @param integer $size The size of the text
3656          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3657          */
3658         function incrementIncludeSize( $type, $size ) {
3659                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3660                         return false;
3661                 } else {
3662                         $this->mIncludeSizes[$type] += $size;
3663                         return true;
3664                 }
3665         }
3666
3667         /**
3668          * Detect __NOGALLERY__ magic word and set a placeholder
3669          */
3670         function stripNoGallery( &$text ) {
3671                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3672                 # do not add TOC
3673                 $mw = MagicWord::get( 'nogallery' );
3674                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3675         }
3676
3677         /**
3678          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3679          * placeholder that will then be replaced by the real TOC in
3680          * ->formatHeadings, this works because at this points real
3681          * comments will have already been discarded by the sanitizer.
3682          *
3683          * Any additional __TOC__ magic words left over will be discarded
3684          * as there can only be one TOC on the page.
3685          */
3686         function stripToc( $text ) {
3687                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3688                 # do not add TOC
3689                 $mw = MagicWord::get( 'notoc' );
3690                 if( $mw->matchAndRemove( $text ) ) {
3691                         $this->mShowToc = false;
3692                 }
3693
3694                 $mw = MagicWord::get( 'toc' );
3695                 if( $mw->match( $text ) ) {
3696                         $this->mShowToc = true;
3697                         $this->mForceTocPosition = true;
3698
3699                         // Set a placeholder. At the end we'll fill it in with the TOC.
3700                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3701
3702                         // Only keep the first one.
3703                         $text = $mw->replace( '', $text );
3704                 }
3705                 return $text;
3706         }
3707
3708         /**
3709          * This function accomplishes several tasks:
3710          * 1) Auto-number headings if that option is enabled
3711          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3712          * 3) Add a Table of contents on the top for users who have enabled the option
3713          * 4) Auto-anchor headings
3714          *
3715          * It loops through all headlines, collects the necessary data, then splits up the
3716          * string and re-inserts the newly formatted headlines.
3717          *
3718          * @param string $text
3719          * @param boolean $isMain
3720          * @private
3721          */
3722         function formatHeadings( $text, $isMain=true ) {
3723                 global $wgMaxTocLevel, $wgContLang;
3724
3725                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3726                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3727                         $showEditLink = 0;
3728                 } else {
3729                         $showEditLink = $this->mOptions->getEditSection();
3730                 }
3731
3732                 # Inhibit editsection links if requested in the page
3733                 $esw =& MagicWord::get( 'noeditsection' );
3734                 if( $esw->matchAndRemove( $text ) ) {
3735                         $showEditLink = 0;
3736                 }
3737
3738                 # Get all headlines for numbering them and adding funky stuff like [edit]
3739                 # links - this is for later, but we need the number of headlines right now
3740                 $matches = array();
3741                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3742
3743                 # if there are fewer than 4 headlines in the article, do not show TOC
3744                 # unless it's been explicitly enabled.
3745                 $enoughToc = $this->mShowToc &&
3746                         (($numMatches >= 4) || $this->mForceTocPosition);
3747
3748                 # Allow user to stipulate that a page should have a "new section"
3749                 # link added via __NEWSECTIONLINK__
3750                 $mw =& MagicWord::get( 'newsectionlink' );
3751                 if( $mw->matchAndRemove( $text ) )
3752                         $this->mOutput->setNewSection( true );
3753
3754                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3755                 # override above conditions and always show TOC above first header
3756                 $mw =& MagicWord::get( 'forcetoc' );
3757                 if ($mw->matchAndRemove( $text ) ) {
3758                         $this->mShowToc = true;
3759                         $enoughToc = true;
3760                 }
3761
3762                 # We need this to perform operations on the HTML
3763                 $sk = $this->mOptions->getSkin();
3764
3765                 # headline counter
3766                 $headlineCount = 0;
3767                 $numVisible = 0;
3768
3769                 # Ugh .. the TOC should have neat indentation levels which can be
3770                 # passed to the skin functions. These are determined here
3771                 $toc = '';
3772                 $full = '';
3773                 $head = array();
3774                 $sublevelCount = array();
3775                 $levelCount = array();
3776                 $toclevel = 0;
3777                 $level = 0;
3778                 $prevlevel = 0;
3779                 $toclevel = 0;
3780                 $prevtoclevel = 0;
3781                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3782                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3783
3784                 foreach( $matches[3] as $headline ) {
3785                         $isTemplate = false;
3786                         $titleText = false;
3787                         $sectionIndex = false;
3788                         $numbering = '';
3789                         $markerMatches = array();
3790                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3791                                 $serial = $markerMatches[1];
3792                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3793                                 $isTemplate = ($titleText != $baseTitleText);
3794                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3795                         }
3796
3797                         if( $toclevel ) {
3798                                 $prevlevel = $level;
3799                                 $prevtoclevel = $toclevel;
3800                         }
3801                         $level = $matches[1][$headlineCount];
3802
3803                         if( $doNumberHeadings || $enoughToc ) {
3804
3805                                 if ( $level > $prevlevel ) {
3806                                         # Increase TOC level
3807                                         $toclevel++;
3808                                         $sublevelCount[$toclevel] = 0;
3809                                         if( $toclevel<$wgMaxTocLevel ) {
3810                                                 $prevtoclevel = $toclevel;
3811                                                 $toc .= $sk->tocIndent();
3812                                                 $numVisible++;
3813                                         }
3814                                 }
3815                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3816                                         # Decrease TOC level, find level to jump to
3817
3818                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3819                                                 # Can only go down to level 1
3820                                                 $toclevel = 1;
3821                                         } else {
3822                                                 for ($i = $toclevel; $i > 0; $i--) {
3823                                                         if ( $levelCount[$i] == $level ) {
3824                                                                 # Found last matching level
3825                                                                 $toclevel = $i;
3826                                                                 break;
3827                                                         }
3828                                                         elseif ( $levelCount[$i] < $level ) {
3829                                                                 # Found first matching level below current level
3830                                                                 $toclevel = $i + 1;
3831                                                                 break;
3832                                                         }
3833                                                 }
3834                                         }
3835                                         if( $toclevel<$wgMaxTocLevel ) {
3836                                                 if($prevtoclevel < $wgMaxTocLevel) {
3837                                                         # Unindent only if the previous toc level was shown :p
3838                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3839                                                 } else {
3840                                                         $toc .= $sk->tocLineEnd();
3841                                                 }
3842                                         }
3843                                 }
3844                                 else {
3845                                         # No change in level, end TOC line
3846                                         if( $toclevel<$wgMaxTocLevel ) {
3847                                                 $toc .= $sk->tocLineEnd();
3848                                         }
3849                                 }
3850
3851                                 $levelCount[$toclevel] = $level;
3852
3853                                 # count number of headlines for each level
3854                                 @$sublevelCount[$toclevel]++;
3855                                 $dot = 0;
3856                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3857                                         if( !empty( $sublevelCount[$i] ) ) {
3858                                                 if( $dot ) {
3859                                                         $numbering .= '.';
3860                                                 }
3861                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3862                                                 $dot = 1;
3863                                         }
3864                                 }
3865                         }
3866
3867                         # The safe header is a version of the header text safe to use for links
3868                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3869                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3870
3871                         # Remove link placeholders by the link text.
3872                         #     <!--LINK number-->
3873                         # turns into
3874                         #     link text with suffix
3875                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3876                                                             "\$this->mLinkHolders['texts'][\$1]",
3877                                                             $safeHeadline );
3878                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3879                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3880                                                             $safeHeadline );
3881
3882                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3883                         $tocline = preg_replace(
3884                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3885                                 array( '',                          '<$1>'),
3886                                 $safeHeadline
3887                         );
3888                         $tocline = trim( $tocline );
3889
3890                         # For the anchor, strip out HTML-y stuff period
3891                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3892                         $safeHeadline = trim( $safeHeadline );
3893
3894                         # Save headline for section edit hint before it's escaped
3895                         $headlineHint = $safeHeadline;
3896                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3897                         $refers[$headlineCount] = $safeHeadline;
3898
3899                         # count how many in assoc. array so we can track dupes in anchors
3900                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3901                         $refcount[$headlineCount] = $refers[$safeHeadline];
3902
3903                         # Don't number the heading if it is the only one (looks silly)
3904                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3905                                 # the two are different if the line contains a link
3906                                 $headline=$numbering . ' ' . $headline;
3907                         }
3908
3909                         # Create the anchor for linking from the TOC to the section
3910                         $anchor = $safeHeadline;
3911                         if($refcount[$headlineCount] > 1 ) {
3912                                 $anchor .= '_' . $refcount[$headlineCount];
3913                         }
3914                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3915                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3916                         }
3917                         # give headline the correct <h#> tag
3918                         if( $showEditLink && $sectionIndex !== false ) {
3919                                 if( $isTemplate )
3920                                         $editlink = $sk->editSectionLinkForOther($titleText, $sectionIndex);
3921                                 else
3922                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3923                         } else {
3924                                 $editlink = '';
3925                         }
3926                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3927
3928                         $headlineCount++;
3929                 }
3930
3931                 # Never ever show TOC if no headers
3932                 if( $numVisible < 1 ) {
3933                         $enoughToc = false;
3934                 }
3935
3936                 if( $enoughToc ) {
3937                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3938                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3939                         }
3940                         $toc = $sk->tocList( $toc );
3941                 }
3942
3943                 # split up and insert constructed headlines
3944
3945                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3946                 $i = 0;
3947
3948                 foreach( $blocks as $block ) {
3949                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3950                                 # This is the [edit] link that appears for the top block of text when
3951                                 # section editing is enabled
3952
3953                                 # Disabled because it broke block formatting
3954                                 # For example, a bullet point in the top line
3955                                 # $full .= $sk->editSectionLink(0);
3956                         }
3957                         $full .= $block;
3958                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3959                                 # Top anchor now in skin
3960                                 $full = $full.$toc;
3961                         }
3962
3963                         if( !empty( $head[$i] ) ) {
3964                                 $full .= $head[$i];
3965                         }
3966                         $i++;
3967                 }
3968                 if( $this->mForceTocPosition ) {
3969                         return str_replace( '<!--MWTOC-->', $toc, $full );
3970                 } else {
3971                         return $full;
3972                 }
3973         }
3974
3975         /**
3976          * Transform wiki markup when saving a page by doing \r\n -> \n
3977          * conversion, substitting signatures, {{subst:}} templates, etc.
3978          *
3979          * @param string $text the text to transform
3980          * @param Title &$title the Title object for the current article
3981          * @param User &$user the User object describing the current user
3982          * @param ParserOptions $options parsing options
3983          * @param bool $clearState whether to clear the parser state first
3984          * @return string the altered wiki markup
3985          * @public
3986          */
3987         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3988                 $this->mOptions = $options;
3989                 $this->mTitle =& $title;
3990                 $this->setOutputType( OT_WIKI );
3991
3992                 if ( $clearState ) {
3993                         $this->clearState();
3994                 }
3995
3996                 $pairs = array(
3997                         "\r\n" => "\n",
3998                 );
3999                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4000                 $text = $this->pstPass2( $text, $user );
4001                 $text = $this->mStripState->unstripBoth( $text );
4002                 return $text;
4003         }
4004
4005         /**
4006          * Pre-save transform helper function
4007          * @private
4008          */
4009         function pstPass2( $text, $user ) {
4010                 global $wgContLang, $wgLocaltimezone;
4011
4012                 /* Note: This is the timestamp saved as hardcoded wikitext to
4013                  * the database, we use $wgContLang here in order to give
4014                  * everyone the same signature and use the default one rather
4015                  * than the one selected in each user's preferences.
4016                  */
4017                 if ( isset( $wgLocaltimezone ) ) {
4018                         $oldtz = getenv( 'TZ' );
4019                         putenv( 'TZ='.$wgLocaltimezone );
4020                 }
4021                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4022                   ' (' . date( 'T' ) . ')';
4023                 if ( isset( $wgLocaltimezone ) ) {
4024                         putenv( 'TZ='.$oldtz );
4025                 }
4026
4027                 # Variable replacement
4028                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4029                 $text = $this->replaceVariables( $text );
4030
4031                 # Strip out <nowiki> etc. added via replaceVariables
4032                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4033
4034                 # Signatures
4035                 $sigText = $this->getUserSig( $user );
4036                 $text = strtr( $text, array(
4037                         '~~~~~' => $d,
4038                         '~~~~' => "$sigText $d",
4039                         '~~~' => $sigText
4040                 ) );
4041
4042                 # Context links: [[|name]] and [[name (context)|]]
4043                 #
4044                 global $wgLegalTitleChars;
4045                 $tc = "[$wgLegalTitleChars]";
4046                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4047
4048                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4049                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4050                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4051
4052                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4053                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4054                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4055
4056                 $t = $this->mTitle->getText();
4057                 $m = array();
4058                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4059                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4060                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4061                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4062                 } else {
4063                         # if there's no context, don't bother duplicating the title
4064                         $text = preg_replace( $p2, '[[\\1]]', $text );
4065                 }
4066
4067                 # Trim trailing whitespace
4068                 $text = rtrim( $text );
4069
4070                 return $text;
4071         }
4072
4073         /**
4074          * Fetch the user's signature text, if any, and normalize to
4075          * validated, ready-to-insert wikitext.
4076          *
4077          * @param User $user
4078          * @return string
4079          * @private
4080          */
4081         function getUserSig( &$user ) {
4082                 global $wgMaxSigChars;
4083
4084                 $username = $user->getName();
4085                 $nickname = $user->getOption( 'nickname' );
4086                 $nickname = $nickname === '' ? $username : $nickname;
4087
4088                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4089                         $nickname = $username;
4090                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4091                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4092                         # Sig. might contain markup; validate this
4093                         if( $this->validateSig( $nickname ) !== false ) {
4094                                 # Validated; clean up (if needed) and return it
4095                                 return $this->cleanSig( $nickname, true );
4096                         } else {
4097                                 # Failed to validate; fall back to the default
4098                                 $nickname = $username;
4099                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4100                         }
4101                 }
4102
4103                 // Make sure nickname doesnt get a sig in a sig
4104                 $nickname = $this->cleanSigInSig( $nickname );
4105
4106                 # If we're still here, make it a link to the user page
4107                 $userText = wfEscapeWikiText( $username );
4108                 $nickText = wfEscapeWikiText( $nickname );
4109                 if ( $user->isAnon() )  {
4110                         return wfMsgForContent( 'signature-anon', $userText, $nickText );
4111                 } else {
4112                         return wfMsgForContent( 'signature', $userText, $nickText );
4113                 }
4114         }
4115
4116         /**
4117          * Check that the user's signature contains no bad XML
4118          *
4119          * @param string $text
4120          * @return mixed An expanded string, or false if invalid.
4121          */
4122         function validateSig( $text ) {
4123                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4124         }
4125
4126         /**
4127          * Clean up signature text
4128          *
4129          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4130          * 2) Substitute all transclusions
4131          *
4132          * @param string $text
4133          * @param $parsing Whether we're cleaning (preferences save) or parsing
4134          * @return string Signature text
4135          */
4136         function cleanSig( $text, $parsing = false ) {
4137                 global $wgTitle;
4138                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
4139
4140                 $substWord = MagicWord::get( 'subst' );
4141                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4142                 $substText = '{{' . $substWord->getSynonym( 0 );
4143
4144                 $text = preg_replace( $substRegex, $substText, $text );
4145                 $text = $this->cleanSigInSig( $text );
4146                 $text = $this->replaceVariables( $text );
4147
4148                 $this->clearState();
4149                 return $text;
4150         }
4151
4152         /**
4153          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4154          * @param string $text
4155          * @return string Signature text with /~{3,5}/ removed
4156          */
4157         function cleanSigInSig( $text ) {
4158                 $text = preg_replace( '/~{3,5}/', '', $text );
4159                 return $text;
4160         }
4161
4162         /**
4163          * Set up some variables which are usually set up in parse()
4164          * so that an external function can call some class members with confidence
4165          * @public
4166          */
4167         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4168                 $this->mTitle =& $title;
4169                 $this->mOptions = $options;
4170                 $this->setOutputType( $outputType );
4171                 if ( $clearState ) {
4172                         $this->clearState();
4173                 }
4174         }
4175
4176         /**
4177          * Transform a MediaWiki message by replacing magic variables.
4178          *
4179          * For some unknown reason, it also expands templates, but only to the
4180          * first recursion level. This is wrong and broken, probably introduced
4181          * accidentally during refactoring, but probably relied upon by thousands
4182          * of users.
4183          *
4184          * @param string $text the text to transform
4185          * @param ParserOptions $options  options
4186          * @return string the text with variables substituted
4187          * @public
4188          */
4189         function transformMsg( $text, $options ) {
4190                 global $wgTitle;
4191                 static $executing = false;
4192
4193                 $fname = "Parser::transformMsg";
4194
4195                 # Guard against infinite recursion
4196                 if ( $executing ) {
4197                         return $text;
4198                 }
4199                 $executing = true;
4200
4201                 wfProfileIn($fname);
4202
4203                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4204                         $this->mTitle = $wgTitle;
4205                 } else {
4206                         $this->mTitle = Title::newFromText('msg');
4207                 }
4208                 $this->mOptions = $options;
4209                 $this->setOutputType( OT_MSG );
4210                 $this->clearState();
4211                 $text = $this->replaceVariables( $text );
4212
4213                 $executing = false;
4214                 wfProfileOut($fname);
4215                 return $text;
4216         }
4217
4218         /**
4219          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4220          * The callback should have the following form:
4221          *    function myParserHook( $text, $params, &$parser ) { ... }
4222          *
4223          * Transform and return $text. Use $parser for any required context, e.g. use
4224          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4225          *
4226          * @public
4227          *
4228          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4229          * @param mixed $callback The callback function (and object) to use for the tag
4230          *
4231          * @return The old value of the mTagHooks array associated with the hook
4232          */
4233         function setHook( $tag, $callback ) {
4234                 $tag = strtolower( $tag );
4235                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4236                 $this->mTagHooks[$tag] = $callback;
4237                 $this->mStripList[] = $tag;
4238
4239                 return $oldVal;
4240         }
4241
4242         function setTransparentTagHook( $tag, $callback ) {
4243                 $tag = strtolower( $tag );
4244                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4245                 $this->mTransparentTagHooks[$tag] = $callback;
4246
4247                 return $oldVal;
4248         }
4249
4250         /**
4251          * Create a function, e.g. {{sum:1|2|3}}
4252          * The callback function should have the form:
4253          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4254          *
4255          * The callback may either return the text result of the function, or an array with the text
4256          * in element 0, and a number of flags in the other elements. The names of the flags are
4257          * specified in the keys. Valid flags are:
4258          *   found                     The text returned is valid, stop processing the template. This
4259          *                             is on by default.
4260          *   nowiki                    Wiki markup in the return value should be escaped
4261          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4262          *   noargs                    Don't replace triple-brace arguments in the return value
4263          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4264          *
4265          * @public
4266          *
4267          * @param string $id The magic word ID
4268          * @param mixed $callback The callback function (and object) to use
4269          * @param integer $flags a combination of the following flags:
4270          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4271          *
4272          * @return The old callback function for this name, if any
4273          */
4274         function setFunctionHook( $id, $callback, $flags = 0 ) {
4275                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4276                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4277
4278                 # Add to function cache
4279                 $mw = MagicWord::get( $id );
4280                 if( !$mw )
4281                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4282
4283                 $synonyms = $mw->getSynonyms();
4284                 $sensitive = intval( $mw->isCaseSensitive() );
4285
4286                 foreach ( $synonyms as $syn ) {
4287                         # Case
4288                         if ( !$sensitive ) {
4289                                 $syn = strtolower( $syn );
4290                         }
4291                         # Add leading hash
4292                         if ( !( $flags & SFH_NO_HASH ) ) {
4293                                 $syn = '#' . $syn;
4294                         }
4295                         # Remove trailing colon
4296                         if ( substr( $syn, -1, 1 ) == ':' ) {
4297                                 $syn = substr( $syn, 0, -1 );
4298                         }
4299                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4300                 }
4301                 return $oldVal;
4302         }
4303
4304         /**
4305          * Get all registered function hook identifiers
4306          *
4307          * @return array
4308          */
4309         function getFunctionHooks() {
4310                 return array_keys( $this->mFunctionHooks );
4311         }
4312
4313         /**
4314          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4315          * Placeholders created in Skin::makeLinkObj()
4316          * Returns an array of links found, indexed by PDBK:
4317          *  0 - broken
4318          *  1 - normal link
4319          *  2 - stub
4320          * $options is a bit field, RLH_FOR_UPDATE to select for update
4321          */
4322         function replaceLinkHolders( &$text, $options = 0 ) {
4323                 global $wgUser;
4324                 global $wgContLang;
4325
4326                 $fname = 'Parser::replaceLinkHolders';
4327                 wfProfileIn( $fname );
4328
4329                 $pdbks = array();
4330                 $colours = 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] = 1;
4360                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4361                                         $colours[$pdbk] = 1;
4362                                         $this->mOutput->addLink( $title, $id );
4363                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4364                                         $colours[$pdbk] = 0;
4365                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4366                                         $colours[$pdbk] = 0;
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                                 # 1 = known
4397                                 # 2 = stub
4398                                 while ( $s = $dbr->fetchObject($res) ) {
4399                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4400                                         $pdbk = $title->getPrefixedDBkey();
4401                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4402                                         $this->mOutput->addLink( $title, $s->page_id );
4403
4404                                         $colours[$pdbk] = ( $threshold == 0 || (
4405                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4406                                                                 $s->page_is_redirect ||
4407                                                                 !Namespace::isContent( $s->page_namespace ) )
4408                                                             ? 1 : 2 );
4409                                 }
4410                         }
4411                         wfProfileOut( $fname.'-check' );
4412
4413                         # Do a second query for different language variants of links and categories
4414                         if($wgContLang->hasVariants()){
4415                                 $linkBatch = new LinkBatch();
4416                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4417                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4418                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4419
4420                                 $categories = $this->mOutput->getCategoryLinks();
4421
4422                                 // Add variants of links to link batch
4423                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4424                                         $title = $this->mLinkHolders['titles'][$key];
4425                                         if ( is_null( $title ) )
4426                                                 continue;
4427
4428                                         $pdbk = $title->getPrefixedDBkey();
4429                                         $titleText = $title->getText();
4430
4431                                         // generate all variants of the link title text
4432                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4433
4434                                         // if link was not found (in first query), add all variants to query
4435                                         if ( !isset($colours[$pdbk]) ){
4436                                                 foreach($allTextVariants as $textVariant){
4437                                                         if($textVariant != $titleText){
4438                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4439                                                                 if(is_null($variantTitle)) continue;
4440                                                                 $linkBatch->addObj( $variantTitle );
4441                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4442                                                         }
4443                                                 }
4444                                         }
4445                                 }
4446
4447                                 // process categories, check if a category exists in some variant
4448                                 foreach( $categories as $category ){
4449                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4450                                         foreach($variants as $variant){
4451                                                 if($variant != $category){
4452                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4453                                                         if(is_null($variantTitle)) continue;
4454                                                         $linkBatch->addObj( $variantTitle );
4455                                                         $categoryMap[$variant] = $category;
4456                                                 }
4457                                         }
4458                                 }
4459
4460
4461                                 if(!$linkBatch->isEmpty()){
4462                                         // construct query
4463                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4464
4465                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4466                                         if ( $threshold > 0 ) {
4467                                                 $variantQuery .= ', page_len, page_is_redirect';
4468                                         }
4469
4470                                         $variantQuery .= " FROM $page WHERE $titleClause";
4471                                         if ( $options & RLH_FOR_UPDATE ) {
4472                                                 $variantQuery .= ' FOR UPDATE';
4473                                         }
4474
4475                                         $varRes = $dbr->query( $variantQuery, $fname );
4476
4477                                         // for each found variants, figure out link holders and replace
4478                                         while ( $s = $dbr->fetchObject($varRes) ) {
4479
4480                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4481                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4482                                                 $vardbk = $variantTitle->getDBkey();
4483
4484                                                 $holderKeys = array();
4485                                                 if(isset($variantMap[$varPdbk])){
4486                                                         $holderKeys = $variantMap[$varPdbk];
4487                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4488                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4489                                                 }
4490
4491                                                 // loop over link holders
4492                                                 foreach($holderKeys as $key){
4493                                                         $title = $this->mLinkHolders['titles'][$key];
4494                                                         if ( is_null( $title ) ) continue;
4495
4496                                                         $pdbk = $title->getPrefixedDBkey();
4497
4498                                                         if(!isset($colours[$pdbk])){
4499                                                                 // found link in some of the variants, replace the link holder data
4500                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4501                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4502
4503                                                                 // set pdbk and colour
4504                                                                 $pdbks[$key] = $varPdbk;
4505                                                                 if ( $threshold >  0 ) {
4506                                                                         $size = $s->page_len;
4507                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4508                                                                                 $colours[$varPdbk] = 1;
4509                                                                         } else {
4510                                                                                 $colours[$varPdbk] = 2;
4511                                                                         }
4512                                                                 }
4513                                                                 else {
4514                                                                         $colours[$varPdbk] = 1;
4515                                                                 }
4516                                                         }
4517                                                 }
4518
4519                                                 // check if the object is a variant of a category
4520                                                 if(isset($categoryMap[$vardbk])){
4521                                                         $oldkey = $categoryMap[$vardbk];
4522                                                         if($oldkey != $vardbk)
4523                                                                 $varCategories[$oldkey]=$vardbk;
4524                                                 }
4525                                         }
4526
4527                                         // rebuild the categories in original order (if there are replacements)
4528                                         if(count($varCategories)>0){
4529                                                 $newCats = array();
4530                                                 $originalCats = $this->mOutput->getCategories();
4531                                                 foreach($originalCats as $cat => $sortkey){
4532                                                         // make the replacement
4533                                                         if( array_key_exists($cat,$varCategories) )
4534                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4535                                                         else $newCats[$cat] = $sortkey;
4536                                                 }
4537                                                 $this->mOutput->setCategoryLinks($newCats);
4538                                         }
4539                                 }
4540                         }
4541
4542                         # Construct search and replace arrays
4543                         wfProfileIn( $fname.'-construct' );
4544                         $replacePairs = array();
4545                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4546                                 $pdbk = $pdbks[$key];
4547                                 $searchkey = "<!--LINK $key-->";
4548                                 $title = $this->mLinkHolders['titles'][$key];
4549                                 if ( empty( $colours[$pdbk] ) ) {
4550                                         $linkCache->addBadLinkObj( $title );
4551                                         $colours[$pdbk] = 0;
4552                                         $this->mOutput->addLink( $title, 0 );
4553                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4554                                                                         $this->mLinkHolders['texts'][$key],
4555                                                                         $this->mLinkHolders['queries'][$key] );
4556                                 } elseif ( $colours[$pdbk] == 1 ) {
4557                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4558                                                                         $this->mLinkHolders['texts'][$key],
4559                                                                         $this->mLinkHolders['queries'][$key] );
4560                                 } elseif ( $colours[$pdbk] == 2 ) {
4561                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4562                                                                         $this->mLinkHolders['texts'][$key],
4563                                                                         $this->mLinkHolders['queries'][$key] );
4564                                 }
4565                         }
4566                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4567                         wfProfileOut( $fname.'-construct' );
4568
4569                         # Do the thing
4570                         wfProfileIn( $fname.'-replace' );
4571                         $text = preg_replace_callback(
4572                                 '/(<!--LINK .*?-->)/',
4573                                 $replacer->cb(),
4574                                 $text);
4575
4576                         wfProfileOut( $fname.'-replace' );
4577                 }
4578
4579                 # Now process interwiki link holders
4580                 # This is quite a bit simpler than internal links
4581                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4582                         wfProfileIn( $fname.'-interwiki' );
4583                         # Make interwiki link HTML
4584                         $replacePairs = array();
4585                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4586                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4587                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4588                         }
4589                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4590
4591                         $text = preg_replace_callback(
4592                                 '/<!--IWLINK (.*?)-->/',
4593                                 $replacer->cb(),
4594                                 $text );
4595                         wfProfileOut( $fname.'-interwiki' );
4596                 }
4597
4598                 wfProfileOut( $fname );
4599                 return $colours;
4600         }
4601
4602         /**
4603          * Replace <!--LINK--> link placeholders with plain text of links
4604          * (not HTML-formatted).
4605          * @param string $text
4606          * @return string
4607          */
4608         function replaceLinkHoldersText( $text ) {
4609                 $fname = 'Parser::replaceLinkHoldersText';
4610                 wfProfileIn( $fname );
4611
4612                 $text = preg_replace_callback(
4613                         '/<!--(LINK|IWLINK) (.*?)-->/',
4614                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4615                         $text );
4616
4617                 wfProfileOut( $fname );
4618                 return $text;
4619         }
4620
4621         /**
4622          * @param array $matches
4623          * @return string
4624          * @private
4625          */
4626         function replaceLinkHoldersTextCallback( $matches ) {
4627                 $type = $matches[1];
4628                 $key  = $matches[2];
4629                 if( $type == 'LINK' ) {
4630                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4631                                 return $this->mLinkHolders['texts'][$key];
4632                         }
4633                 } elseif( $type == 'IWLINK' ) {
4634                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4635                                 return $this->mInterwikiLinkHolders['texts'][$key];
4636                         }
4637                 }
4638                 return $matches[0];
4639         }
4640
4641         /**
4642          * Tag hook handler for 'pre'.
4643          */
4644         function renderPreTag( $text, $attribs ) {
4645                 // Backwards-compatibility hack
4646                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4647
4648                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4649                 return wfOpenElement( 'pre', $attribs ) .
4650                         Xml::escapeTagsOnly( $content ) .
4651                         '</pre>';
4652         }
4653
4654         /**
4655          * Renders an image gallery from a text with one line per image.
4656          * text labels may be given by using |-style alternative text. E.g.
4657          *   Image:one.jpg|The number "1"
4658          *   Image:tree.jpg|A tree
4659          * given as text will return the HTML of a gallery with two images,
4660          * labeled 'The number "1"' and
4661          * 'A tree'.
4662          */
4663         function renderImageGallery( $text, $params ) {
4664                 $ig = new ImageGallery();
4665                 $ig->setContextTitle( $this->mTitle );
4666                 $ig->setShowBytes( false );
4667                 $ig->setShowFilename( false );
4668                 $ig->setParser( $this );
4669                 $ig->setHideBadImages();
4670                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4671                 $ig->useSkin( $this->mOptions->getSkin() );
4672                 $ig->mRevisionId = $this->mRevisionId;
4673
4674                 if( isset( $params['caption'] ) ) {
4675                         $caption = $params['caption'];
4676                         $caption = htmlspecialchars( $caption );
4677                         $caption = $this->replaceInternalLinks( $caption );
4678                         $ig->setCaptionHtml( $caption );
4679                 }
4680                 if( isset( $params['perrow'] ) ) {
4681                         $ig->setPerRow( $params['perrow'] );
4682                 }
4683                 if( isset( $params['widths'] ) ) {
4684                         $ig->setWidths( $params['widths'] );
4685                 }
4686                 if( isset( $params['heights'] ) ) {
4687                         $ig->setHeights( $params['heights'] );
4688                 }
4689
4690                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4691
4692                 $lines = explode( "\n", $text );
4693                 foreach ( $lines as $line ) {
4694                         # match lines like these:
4695                         # Image:someimage.jpg|This is some image
4696                         $matches = array();
4697                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4698                         # Skip empty lines
4699                         if ( count( $matches ) == 0 ) {
4700                                 continue;
4701                         }
4702                         $tp = Title::newFromText( $matches[1] );
4703                         $nt =& $tp;
4704                         if( is_null( $nt ) ) {
4705                                 # Bogus title. Ignore these so we don't bomb out later.
4706                                 continue;
4707                         }
4708                         if ( isset( $matches[3] ) ) {
4709                                 $label = $matches[3];
4710                         } else {
4711                                 $label = '';
4712                         }
4713
4714                         $pout = $this->parse( $label,
4715                                 $this->mTitle,
4716                                 $this->mOptions,
4717                                 false, // Strip whitespace...?
4718                                 false  // Don't clear state!
4719                         );
4720                         $html = $pout->getText();
4721
4722                         $ig->add( $nt, $html );
4723
4724                         # Only add real images (bug #5586)
4725                         if ( $nt->getNamespace() == NS_IMAGE ) {
4726                                 $this->mOutput->addImage( $nt->getDBkey() );
4727                         }
4728                 }
4729                 return $ig->toHTML();
4730         }
4731
4732         function getImageParams( $handler ) {
4733                 if ( $handler ) {
4734                         $handlerClass = get_class( $handler );
4735                 } else {
4736                         $handlerClass = '';
4737                 }
4738                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4739                         // Initialise static lists
4740                         static $internalParamNames = array(
4741                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4742                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4743                                         'bottom', 'text-bottom' ),
4744                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4745                                         'upright', 'border' ),
4746                         );
4747                         static $internalParamMap;
4748                         if ( !$internalParamMap ) {
4749                                 $internalParamMap = array();
4750                                 foreach ( $internalParamNames as $type => $names ) {
4751                                         foreach ( $names as $name ) {
4752                                                 $magicName = str_replace( '-', '_', "img_$name" );
4753                                                 $internalParamMap[$magicName] = array( $type, $name );
4754                                         }
4755                                 }
4756                         }
4757
4758                         // Add handler params
4759                         $paramMap = $internalParamMap;
4760                         if ( $handler ) {
4761                                 $handlerParamMap = $handler->getParamMap();
4762                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4763                                         $paramMap[$magic] = array( 'handler', $paramName );
4764                                 }
4765                         }
4766                         $this->mImageParams[$handlerClass] = $paramMap;
4767                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4768                 }
4769                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4770         }
4771
4772         /**
4773          * Parse image options text and use it to make an image
4774          */
4775         function makeImage( $title, $options ) {
4776                 # @TODO: let the MediaHandler specify its transform parameters
4777                 #
4778                 # Check if the options text is of the form "options|alt text"
4779                 # Options are:
4780                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4781                 #  * left               no resizing, just left align. label is used for alt= only
4782                 #  * right              same, but right aligned
4783                 #  * none               same, but not aligned
4784                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4785                 #  * center             center the image
4786                 #  * framed             Keep original image size, no magnify-button.
4787                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4788                 #  * upright            reduce width for upright images, rounded to full __0 px
4789                 #  * border             draw a 1px border around the image
4790                 # vertical-align values (no % or length right now):
4791                 #  * baseline
4792                 #  * sub
4793                 #  * super
4794                 #  * top
4795                 #  * text-top
4796                 #  * middle
4797                 #  * bottom
4798                 #  * text-bottom
4799
4800                 $parts = array_map( 'trim', explode( '|', $options) );
4801                 $sk = $this->mOptions->getSkin();
4802
4803                 # Give extensions a chance to select the file revision for us
4804                 $skip = $time = false;
4805                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4806
4807                 if ( $skip ) {
4808                         return $sk->makeLinkObj( $title );
4809                 }
4810
4811                 # Get parameter map
4812                 $file = wfFindFile( $title, $time );
4813                 $handler = $file ? $file->getHandler() : false;
4814
4815                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4816
4817                 # Process the input parameters
4818                 $caption = '';
4819                 $params = array( 'frame' => array(), 'handler' => array(),
4820                         'horizAlign' => array(), 'vertAlign' => array() );
4821                 foreach( $parts as $part ) {
4822                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4823                         if ( isset( $paramMap[$magicName] ) ) {
4824                                 list( $type, $paramName ) = $paramMap[$magicName];
4825                                 $params[$type][$paramName] = $value;
4826
4827                                 // Special case; width and height come in one variable together
4828                                 if( $type == 'handler' && $paramName == 'width' ) {
4829                                         $m = array();
4830                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4831                                                 $params[$type]['width'] = intval( $m[1] );
4832                                                 $params[$type]['height'] = intval( $m[2] );
4833                                         } else {
4834                                                 $params[$type]['width'] = intval( $value );
4835                                         }
4836                                 }
4837                         } else {
4838                                 $caption = $part;
4839                         }
4840                 }
4841
4842                 # Process alignment parameters
4843                 if ( $params['horizAlign'] ) {
4844                         $params['frame']['align'] = key( $params['horizAlign'] );
4845                 }
4846                 if ( $params['vertAlign'] ) {
4847                         $params['frame']['valign'] = key( $params['vertAlign'] );
4848                 }
4849
4850                 # Validate the handler parameters
4851                 if ( $handler ) {
4852                         foreach ( $params['handler'] as $name => $value ) {
4853                                 if ( !$handler->validateParam( $name, $value ) ) {
4854                                         unset( $params['handler'][$name] );
4855                                 }
4856                         }
4857                 }
4858
4859                 # Strip bad stuff out of the alt text
4860                 $alt = $this->replaceLinkHoldersText( $caption );
4861
4862                 # make sure there are no placeholders in thumbnail attributes
4863                 # that are later expanded to html- so expand them now and
4864                 # remove the tags
4865                 $alt = $this->mStripState->unstripBoth( $alt );
4866                 $alt = Sanitizer::stripAllTags( $alt );
4867
4868                 $params['frame']['alt'] = $alt;
4869                 $params['frame']['caption'] = $caption;
4870
4871                 # Linker does the rest
4872                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4873
4874                 # Give the handler a chance to modify the parser object
4875                 if ( $handler ) {
4876                         $handler->parserTransformHook( $this, $file );
4877                 }
4878
4879                 return $ret;
4880         }
4881
4882         /**
4883          * Set a flag in the output object indicating that the content is dynamic and
4884          * shouldn't be cached.
4885          */
4886         function disableCache() {
4887                 wfDebug( "Parser output marked as uncacheable.\n" );
4888                 $this->mOutput->mCacheTime = -1;
4889         }
4890
4891         /**#@+
4892          * Callback from the Sanitizer for expanding items found in HTML attribute
4893          * values, so they can be safely tested and escaped.
4894          * @param string $text
4895          * @param PPFrame $frame
4896          * @return string
4897          * @private
4898          */
4899         function attributeStripCallback( &$text, $frame = false ) {
4900                 $text = $this->replaceVariables( $text, $frame );
4901                 $text = $this->mStripState->unstripBoth( $text );
4902                 return $text;
4903         }
4904
4905         /**#@-*/
4906
4907         /**#@+
4908          * Accessor/mutator
4909          */
4910         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4911         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4912         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4913         /**#@-*/
4914
4915         /**#@+
4916          * Accessor
4917          */
4918         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4919         /**#@-*/
4920
4921
4922         /**
4923          * Break wikitext input into sections, and either pull or replace
4924          * some particular section's text.
4925          *
4926          * External callers should use the getSection and replaceSection methods.
4927          *
4928          * @param $text Page wikitext
4929          * @param $section Numbered section. 0 pulls the text before the first
4930          *                 heading; other numbers will pull the given section
4931          *                 along with its lower-level subsections. If the section is
4932          *                 not found, $mode=get will return $newtext, and
4933          *                 $mode=replace will return $text.
4934          * @param $mode One of "get" or "replace"
4935          * @param $newText Replacement text for section data.
4936          * @return string for "get", the extracted section text.
4937          *                for "replace", the whole page with the section replaced.
4938          */
4939         private function extractSections( $text, $section, $mode, $newText='' ) {
4940                 $this->clearState();
4941                 $this->mOptions = new ParserOptions;
4942                 $this->setOutputType( OT_WIKI );
4943                 $curIndex = 0;
4944                 $outText = '';
4945                 $frame = new PPFrame( $this );
4946
4947                 // Preprocess the text
4948                 $dom = $this->preprocessToDom( $text );
4949                 $root = $dom->documentElement;
4950
4951                 // <h> nodes indicate section breaks
4952                 // They can only occur at the top level, so we can find them by iterating the root's children
4953                 $node = $root->firstChild;
4954
4955                 // Find the target section
4956                 if ( $section == 0 ) {
4957                         // Section zero doesn't nest, level=big
4958                         $targetLevel = 1000;
4959                 } else {
4960                         while ( $node ) {
4961                                 if ( $node->nodeName == 'h' ) {
4962                                         if ( $curIndex + 1 == $section ) {
4963                                                 break;
4964                                         }
4965                                         $curIndex++;
4966                                 }
4967                                 if ( $mode == 'replace' ) {
4968                                         $outText .= $frame->expand( $node );
4969                                 }
4970                                 $node = $node->nextSibling;
4971                         }
4972                         if ( $node ) {
4973                                 $targetLevel = $node->getAttribute( 'level' );
4974                         }
4975                 }
4976
4977                 if ( !$node ) {
4978                         // Not found
4979                         if ( $mode == 'get' ) {
4980                                 return $newText;
4981                         } else {
4982                                 return $text;
4983                         }
4984                 }
4985
4986                 // Find the end of the section, including nested sections
4987                 do {
4988                         if ( $node->nodeName == 'h' ) {
4989                                 $curIndex++;
4990                                 $curLevel = $node->getAttribute( 'level' );
4991                                 if ( $curIndex != $section && $curLevel <= $targetLevel ) {
4992                                         break;
4993                                 }
4994                         }
4995                         if ( $mode == 'get' ) {
4996                                 $outText .= $frame->expand( $node );
4997                         }
4998                         $node = $node->nextSibling;
4999                 } while ( $node );
5000
5001                 // Write out the remainder (in replace mode only)
5002                 if ( $mode == 'replace' ) {
5003                         // Output the replacement text
5004                         // Add two newlines on -- trailing whitespace in $newText is conventionally
5005                         // stripped by the editor, so we need both newlines to restore the paragraph gap
5006                         $outText .= $newText . "\n\n";
5007                         while ( $node ) {
5008                                 $outText .= $frame->expand( $node );
5009                                 $node = $node->nextSibling;
5010                         }
5011                 }
5012
5013                 if ( is_string( $outText ) ) {
5014                         // Re-insert stripped tags
5015                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5016                 }
5017
5018                 return $outText;
5019         }
5020
5021         /**
5022          * This function returns the text of a section, specified by a number ($section).
5023          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5024          * the first section before any such heading (section 0).
5025          *
5026          * If a section contains subsections, these are also returned.
5027          *
5028          * @param $text String: text to look in
5029          * @param $section Integer: section number
5030          * @param $deftext: default to return if section is not found
5031          * @return string text of the requested section
5032          */
5033         public function getSection( $text, $section, $deftext='' ) {
5034                 return $this->extractSections( $text, $section, "get", $deftext );
5035         }
5036
5037         public function replaceSection( $oldtext, $section, $text ) {
5038                 return $this->extractSections( $oldtext, $section, "replace", $text );
5039         }
5040
5041         /**
5042          * Get the timestamp associated with the current revision, adjusted for
5043          * the default server-local timestamp
5044          */
5045         function getRevisionTimestamp() {
5046                 if ( is_null( $this->mRevisionTimestamp ) ) {
5047                         wfProfileIn( __METHOD__ );
5048                         global $wgContLang;
5049                         $dbr = wfGetDB( DB_SLAVE );
5050                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5051                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5052
5053                         // Normalize timestamp to internal MW format for timezone processing.
5054                         // This has the added side-effect of replacing a null value with
5055                         // the current time, which gives us more sensible behavior for
5056                         // previews.
5057                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5058
5059                         // The cryptic '' timezone parameter tells to use the site-default
5060                         // timezone offset instead of the user settings.
5061                         //
5062                         // Since this value will be saved into the parser cache, served
5063                         // to other users, and potentially even used inside links and such,
5064                         // it needs to be consistent for all visitors.
5065                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5066
5067                         wfProfileOut( __METHOD__ );
5068                 }
5069                 return $this->mRevisionTimestamp;
5070         }
5071
5072         /**
5073          * Mutator for $mDefaultSort
5074          *
5075          * @param $sort New value
5076          */
5077         public function setDefaultSort( $sort ) {
5078                 $this->mDefaultSort = $sort;
5079         }
5080
5081         /**
5082          * Accessor for $mDefaultSort
5083          * Will use the title/prefixed title if none is set
5084          *
5085          * @return string
5086          */
5087         public function getDefaultSort() {
5088                 if( $this->mDefaultSort !== false ) {
5089                         return $this->mDefaultSort;
5090                 } else {
5091                         return $this->mTitle->getNamespace() == NS_CATEGORY
5092                                         ? $this->mTitle->getText()
5093                                         : $this->mTitle->getPrefixedText();
5094                 }
5095         }
5096
5097         /**
5098          * Try to guess the section anchor name based on a wikitext fragment
5099          * presumably extracted from a heading, for example "Header" from
5100          * "== Header ==".
5101          */
5102         public function guessSectionNameFromWikiText( $text ) {
5103                 # Strip out wikitext links(they break the anchor)
5104                 $text = $this->stripSectionName( $text );
5105                 $headline = Sanitizer::decodeCharReferences( $text );
5106                 # strip out HTML
5107                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5108                 $headline = trim( $headline );
5109                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5110                 $replacearray = array(
5111                         '%3A' => ':',
5112                         '%' => '.'
5113                 );
5114                 return str_replace(
5115                         array_keys( $replacearray ),
5116                         array_values( $replacearray ),
5117                         $sectionanchor );
5118         }
5119
5120         /**
5121          * Strips a text string of wikitext for use in a section anchor
5122          *
5123          * Accepts a text string and then removes all wikitext from the
5124          * string and leaves only the resultant text (i.e. the result of
5125          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5126          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5127          * to create valid section anchors by mimicing the output of the
5128          * parser when headings are parsed.
5129          *
5130          * @param $text string Text string to be stripped of wikitext
5131          * for use in a Section anchor
5132          * @return Filtered text string
5133          */
5134         public function stripSectionName( $text ) {
5135                 # Strip internal link markup
5136                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5137                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5138
5139                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5140                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5141                 # on how many empty links there are on the page - need to figure that out.
5142                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5143
5144                 # Parse wikitext quotes (italics & bold)
5145                 $text = $this->doQuotes($text);
5146
5147                 # Strip HTML tags
5148                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5149                 return $text;
5150         }
5151
5152         /**
5153          * strip/replaceVariables/unstrip for preprocessor regression testing
5154          */
5155         function srvus( $text ) {
5156                 $text = $this->replaceVariables( $text );
5157                 $text = $this->mStripState->unstripBoth( $text );
5158                 return $text;
5159         }
5160 }
5161
5162 /**
5163  * @todo document, briefly.
5164  * @addtogroup Parser
5165  */
5166 class OnlyIncludeReplacer {
5167         var $output = '';
5168
5169         function replace( $matches ) {
5170                 if ( substr( $matches[1], -1 ) == "\n" ) {
5171                         $this->output .= substr( $matches[1], 0, -1 );
5172                 } else {
5173                         $this->output .= $matches[1];
5174                 }
5175         }
5176 }
5177
5178 /**
5179  * @todo document, briefly.
5180  * @addtogroup Parser
5181  */
5182 class StripState {
5183         var $general, $nowiki;
5184
5185         function __construct() {
5186                 $this->general = new ReplacementArray;
5187                 $this->nowiki = new ReplacementArray;
5188         }
5189
5190         function unstripGeneral( $text ) {
5191                 wfProfileIn( __METHOD__ );
5192                 do {
5193                         $oldText = $text;
5194                         $text = $this->general->replace( $text );
5195                 } while ( $text != $oldText );
5196                 wfProfileOut( __METHOD__ );
5197                 return $text;
5198         }
5199
5200         function unstripNoWiki( $text ) {
5201                 wfProfileIn( __METHOD__ );
5202                 do {
5203                         $oldText = $text;
5204                         $text = $this->nowiki->replace( $text );
5205                 } while ( $text != $oldText );
5206                 wfProfileOut( __METHOD__ );
5207                 return $text;
5208         }
5209
5210         function unstripBoth( $text ) {
5211                 wfProfileIn( __METHOD__ );
5212                 do {
5213                         $oldText = $text;
5214                         $text = $this->general->replace( $text );
5215                         $text = $this->nowiki->replace( $text );
5216                 } while ( $text != $oldText );
5217                 wfProfileOut( __METHOD__ );
5218                 return $text;
5219         }
5220 }
5221
5222 /**
5223  * An expansion frame, used as a context to expand the result of preprocessToDom()
5224  */
5225 class PPFrame {
5226         var $parser, $title;
5227
5228         const NO_ARGS = 1;
5229         const NO_TEMPLATES = 2;
5230
5231         /**
5232          * Construct a new preprocessor frame.
5233          * @param Parser $parser The parent parser
5234          * @param Title $title The context title, or false if there isn't one
5235          */
5236         function __construct( $parser ) {
5237                 $this->parser = $parser;
5238                 $this->title = $parser->mTitle;
5239         }
5240
5241         /**
5242          * Create a new child frame
5243          * $args is optionally a DOMNodeList containing the template arguments
5244          */
5245         function newChild( $args = false, $title = false ) {
5246                 $assocArgs = array();
5247                 if ( $title === false ) {
5248                         $title = $this->title;
5249                 }
5250                 if ( $args !== false ) {
5251                         $xpath = false;
5252                         foreach ( $args as $arg ) {
5253                                 if ( !$xpath ) {
5254                                         $xpath = new DOMXPath( $arg->ownerDocument );
5255                                 }
5256
5257                                 $nameNodes = $xpath->query( 'name', $arg );
5258                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5259                                         // Numbered parameter
5260                                         $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5261                                 } else {
5262                                         // Named parameter
5263                                         $name = $this->expand( $nameNodes->item( 0 ) );
5264                                 }
5265
5266                                 $value = $xpath->query( 'value', $arg );
5267                                 $assocArgs[$name] = $value->item( 0 );
5268                         }
5269                 }
5270                 return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
5271         }
5272
5273         /**
5274          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5275          * using the current context
5276          * @param $root the node
5277          */
5278         function expand( $root, $shallowFlags = 0, $deepFlags = 0 ) {
5279                 if ( is_string( $root ) ) {
5280                         return $root;
5281                 }
5282
5283                 if ( $this->parser->ot['html']
5284                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5285                 {
5286                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5287                 }
5288                 $flags = $shallowFlags | $deepFlags;
5289
5290                 if ( is_array( $root ) ) {
5291                         $s = '';
5292                         foreach ( $root as $node ) {
5293                                 $s .= $this->expand( $node, 0, $deepFlags );
5294                         }
5295                 } elseif ( $root instanceof DOMNodeList ) {
5296                         $s = '';
5297                         foreach ( $root as $node ) {
5298                                 $s .= $this->expand( $node, 0, $deepFlags );
5299                         }
5300                 } elseif ( $root instanceof DOMNode ) {
5301                         if ( $root->nodeType == XML_TEXT_NODE ) {
5302                                 $s = $root->nodeValue;
5303                         } elseif ( $root->nodeName == 'template' ) {
5304                                 # Double-brace expansion
5305                                 $xpath = new DOMXPath( $root->ownerDocument );
5306                                 $titles = $xpath->query( 'title', $root );
5307                                 $title = $titles->item( 0 );
5308                                 $parts = $xpath->query( 'part', $root );
5309                                 if ( $flags & self::NO_TEMPLATES ) {
5310                                         $s = '{{' . $this->implodeWithFlags( '|', 0, $deepFlags, $title, $parts ) . '}}';
5311                                 } else {
5312                                         $lineStart = $root->getAttribute( 'lineStart' );
5313                                         $params = array(
5314                                                 'title' => $title,
5315                                                 'parts' => $parts,
5316                                                 'lineStart' => $lineStart,
5317                                                 'text' => 'FIXME' );
5318                                         $s = $this->parser->braceSubstitution( $params, $this );
5319                                 }
5320                         } elseif ( $root->nodeName == 'tplarg' ) {
5321                                 # Triple-brace expansion
5322                                 $xpath = new DOMXPath( $root->ownerDocument );
5323                                 $titles = $xpath->query( 'title', $root );
5324                                 $title = $titles->item( 0 );
5325                                 $parts = $xpath->query( 'part', $root );
5326                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5327                                         $s = '{{{' . $this->implode( '|', 0, $deepFlags, $title, $parts ) . '}}}';
5328                                 } else {
5329                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5330                                         $s = $this->parser->argSubstitution( $params, $this );
5331                                 }
5332                         } elseif ( $root->nodeName == 'ext' ) {
5333                                 # Extension tag
5334                                 $xpath = new DOMXPath( $root->ownerDocument );
5335                                 $names = $xpath->query( 'name', $root );
5336                                 $attrs = $xpath->query( 'attr', $root );
5337                                 $inners = $xpath->query( 'inner', $root );
5338                                 $closes = $xpath->query( 'close', $root );
5339                                 $params = array(
5340                                         'name' => $names->item( 0 ),
5341                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5342                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5343                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5344                                 );
5345                                 $s = $this->parser->extensionSubstitution( $params, $this );
5346                         } elseif ( $root->nodeName == 'h' ) {
5347                                 # Heading
5348                                 $s = $this->expand( $root->childNodes, 0, $deepFlags );
5349
5350                                 if ( $this->parser->ot['html'] ) {
5351                                         # Insert heading index marker
5352                                         $headingIndex = $root->getAttribute( 'i' );
5353                                         $titleText = $this->title->getPrefixedDBkey();
5354                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5355                                         $serial = count( $this->parser->mHeadings ) - 1;
5356                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5357                                         $count = $root->getAttribute( 'level' );
5358
5359                                         // FIXME: bug-for-bug with old parser
5360                                         // Lose whitespace for no apparent reason
5361                                         // Remove this after differential testing is done
5362                                         if ( true ) {
5363                                                 // Good version
5364                                                 $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5365                                         } else {
5366                                                 // Bad version
5367                                                 if ( preg_match( '/^(={1,6})(.*?)(={1,6})\s*?$/', $s, $m ) ) {
5368                                                         if ( $m[2] != '' ) {
5369                                                                 $s = $m[1] . $marker . $m[2] . $m[3];
5370                                                         }
5371                                                 }
5372                                         }
5373                                         $this->parser->mStripState->general->setPair( $marker, '' );
5374                                 }
5375                         } else {
5376                                 # Generic recursive expansion
5377                                 $s = '';
5378                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5379                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5380                                                 $s .= $node->nodeValue;
5381                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5382                                                 $s .= $this->expand( $node, 0, $deepFlags );
5383                                         }
5384                                 }
5385                         }
5386                 } else {
5387                         throw new MWException( __METHOD__.': Invalid parameter type' );
5388                 }
5389                 return $s;
5390         }
5391
5392         function implodeWithFlags( $sep, $shallowFlags, $deepFlags /*, ... */ ) {
5393                 $args = array_slice( func_get_args(), 3 );
5394
5395                 $first = true;
5396                 $s = '';
5397                 foreach ( $args as $root ) {
5398                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5399                                 $root = array( $root );
5400                         }
5401                         foreach ( $root as $node ) {
5402                                 if ( $first ) {
5403                                         $first = false;
5404                                 } else {
5405                                         $s .= $sep;
5406                                 }
5407                                 $s .= $this->expand( $node, $shallowFlags, $deepFlags );
5408                         }
5409                 }
5410                 return $s;
5411         }
5412
5413         function implode( $sep /*, ... */ ) {
5414                 $args = func_get_args();
5415                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0, 0 ), array_slice( $args, 1 ) );
5416                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5417         }
5418
5419         function __toString() {
5420                 return 'frame{}';
5421         }
5422 }
5423
5424 /**
5425  * Expansion frame with template arguments
5426  */
5427 class PPTemplateFrame extends PPFrame {
5428         public $parser, $args, $parent, $serial;
5429
5430         function __construct( $parser, $parent = false, $args = array(), $title = false ) {
5431                 $this->parser = $parser;
5432                 $this->parent = $parent;
5433                 $this->args = $args;
5434                 $this->title = $title;
5435         }
5436
5437         function __toString() {
5438                 $s = 'tplframe{';
5439                 $first = true;
5440                 foreach ( $this->args as $name => $value ) {
5441                         if ( $first ) {
5442                                 $first = false;
5443                         } else {
5444                                 $s .= ', ';
5445                         }
5446                         $s .= "\"$name\":\"" .
5447                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5448                 }
5449                 $s .= '}';
5450                 return $s;
5451         }
5452 }
5453