includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are four main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * transformMsg()
  24  *   performs brace substitution on MediaWiki messages
  25  * preprocess()
  26  *   removes HTML comments and expands templates
  27  *
  28  * Globals used:
  29  *    objects:   $wgLang, $wgContLang
  30  *
  31  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  32  *
  33  * settings:
  34  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  35  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  36  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  37  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.4';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  62                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  63
  64         // State constants for the definition list colon extraction
  65         const COLON_STATE_TEXT = 0;
  66         const COLON_STATE_TAG = 1;
  67         const COLON_STATE_TAGSTART = 2;
  68         const COLON_STATE_CLOSETAG = 3;
  69         const COLON_STATE_TAGSLASH = 4;
  70         const COLON_STATE_COMMENT = 5;
  71         const COLON_STATE_COMMENTDASH = 6;
  72         const COLON_STATE_COMMENTDASHDASH = 7;
  73
  74         // Flags for preprocessToDom
  75         const PTD_FOR_INCLUSION = 1;
  76
  77         /**#@+
  78          * @private
  79          */
  80         # Persistent:
  81         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  82                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  83                 $mExtLinkBracketedRegex;
  84
  85         # Cleared with clearState():
  86         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  87         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  88         var $mInterwikiLinkHolders, $mLinkHolders;
  89         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  90         var $mTplExpandCache,// empty-frame expansion cache
  91             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  92                                 // in this path. Used for loop detection.
  93         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  94
  95         # Temporary
  96         # These are variables reset at least once per parse regardless of $clearState
  97         var $mOptions,      // ParserOptions object
  98                 $mTitle,        // Title context, used for self-link rendering and similar things
  99                 $mOutputType,   // Output type, one of the OT_xxx constants
 100                 $ot,            // Shortcut alias, see setOutputType()
 101                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 102                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 103                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 104
 105         /**#@-*/
 106
 107         /**
 108          * Constructor
 109          *
 110          * @public
 111          */
 112         function __construct( $conf = array() ) {
 113                 $this->mTagHooks = array();
 114                 $this->mTransparentTagHooks = array();
 115                 $this->mFunctionHooks = array();
 116                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 117                 $this->mStripList = array( 'nowiki', 'gallery' );
 118                 $this->mMarkerSuffix = "-QINU\x7f";
 119                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 120                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 121                 $this->mFirstCall = true;
 122         }
 123
 124         /**
 125          * Do various kinds of initialisation on the first call of the parser
 126          */
 127         function firstCallInit() {
 128                 if ( !$this->mFirstCall ) {
 129                         return;
 130                 }
 131
 132                 wfProfileIn( __METHOD__ );
 133                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 134
 135                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 136
 137                 # Syntax for arguments (see self::setFunctionHook):
 138                 #  "name for lookup in localized magic words array",
 139                 #  function callback,
 140                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 141                 #    instead of {{#int:...}})
 142                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 167                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'tag',              array( 'CoreParserFunctions', 'tagObj'           ), SFH_OBJECT_ARGS );
 170
 171                 if ( $wgAllowDisplayTitle ) {
 172                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 173                 }
 174                 if ( $wgAllowSlowParserFunctions ) {
 175                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 176                 }
 177
 178                 $this->initialiseVariables();
 179                 $this->mFirstCall = false;
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = false;
 200                 $this->mInPre = false;
 201                 $this->mInterwikiLinkHolders = array(
 202                         'texts' => array(),
 203                         'titles' => array()
 204                 );
 205                 $this->mLinkHolders = array(
 206                         'namespaces' => array(),
 207                         'dbkeys' => array(),
 208                         'queries' => array(),
 209                         'texts' => array(),
 210                         'titles' => array()
 211                 );
 212                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 213
 214                 /**
 215                  * Prefix for temporary replacement strings for the multipass parser.
 216                  * \x07 should never appear in input as it's disallowed in XML.
 217                  * Using it at the front also gives us a little extra robustness
 218                  * since it shouldn't match when butted up against identifier-like
 219                  * string constructs.
 220                  *
 221                  * Must not consist of all title characters, or else it will change
 222                  * the behaviour of <nowiki> in a link.
 223                  */
 224                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 225                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 226
 227                 # Clear these on every parse, bug 4549
 228                 $this->mTemplatePath = array();
 229                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 230
 231                 $this->mShowToc = true;
 232                 $this->mForceTocPosition = false;
 233                 $this->mIncludeSizes = array(
 234                         'post-expand' => 0,
 235                         'arg' => 0,
 236                 );
 237                 $this->mPPNodeCount = 0;
 238                 $this->mDefaultSort = false;
 239                 $this->mHeadings = array();
 240
 241                 wfRunHooks( 'ParserClearState', array( &$this ) );
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         function setOutputType( $ot ) {
 246                 $this->mOutputType = $ot;
 247                 // Shortcut alias
 248                 $this->ot = array(
 249                         'html' => $ot == OT_HTML,
 250                         'wiki' => $ot == OT_WIKI,
 251                         'msg' => $ot == OT_MSG,
 252                         'pre' => $ot == OT_PREPROCESS,
 253                 );
 254         }
 255
 256         /**
 257          * Accessor for mUniqPrefix.
 258          *
 259          * @public
 260          */
 261         function uniqPrefix() {
 262                 if( !isset( $this->mUniqPrefix ) ) {
 263                         // @fixme this is probably *horribly wrong*
 264                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 265                         // if this is called for a parser cache hit, the parser may not
 266                         // have ever been initialized in the first place.
 267                         // Not really sure what the heck is supposed to be going on here.
 268                         return '';
 269                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 270                 }
 271                 return $this->mUniqPrefix;
 272         }
 273
 274         /**
 275          * Convert wikitext to HTML
 276          * Do not call this function recursively.
 277          *
 278          * @param string $text Text we want to parse
 279          * @param Title &$title A title object
 280          * @param array $options
 281          * @param boolean $linestart
 282          * @param boolean $clearState
 283          * @param int $revid number to pass in {{REVISIONID}}
 284          * @return ParserOutput a ParserOutput
 285          */
 286         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 287                 /**
 288                  * First pass--just handle <nowiki> sections, pass the rest off
 289                  * to internalParse() which does all the real work.
 290                  */
 291
 292                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 293                 $fname = 'Parser::parse-' . wfGetCaller();
 294                 wfProfileIn( __METHOD__ );
 295                 wfProfileIn( $fname );
 296
 297                 if ( $clearState ) {
 298                         $this->clearState();
 299                 }
 300
 301                 $this->mOptions = $options;
 302                 $this->mTitle =& $title;
 303                 $oldRevisionId = $this->mRevisionId;
 304                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 305                 if( $revid !== null ) {
 306                         $this->mRevisionId = $revid;
 307                         $this->mRevisionTimestamp = null;
 308                 }
 309                 $this->setOutputType( OT_HTML );
 310                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 311                 # No more strip!
 312                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 313                 $text = $this->internalParse( $text );
 314                 $text = $this->mStripState->unstripGeneral( $text );
 315
 316                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 317                 $fixtags = array(
 318                         # french spaces, last one Guillemet-left
 319                         # only if there is something before the space
 320                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 321                         # french spaces, Guillemet-right
 322                         '/(\\302\\253) /' => '\\1&nbsp;',
 323                 );
 324                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 325
 326                 # only once and last
 327                 $text = $this->doBlockLevels( $text, $linestart );
 328
 329                 $this->replaceLinkHolders( $text );
 330
 331                 # the position of the parserConvert() call should not be changed. it
 332                 # assumes that the links are all replaced and the only thing left
 333                 # is the <nowiki> mark.
 334                 # Side-effects: this calls $this->mOutput->setTitleText()
 335                 $text = $wgContLang->parserConvert( $text, $this );
 336
 337                 $text = $this->mStripState->unstripNoWiki( $text );
 338
 339                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 340
 341 //!JF Move to its own function
 342
 343                 $uniq_prefix = $this->mUniqPrefix;
 344                 $matches = array();
 345                 $elements = array_keys( $this->mTransparentTagHooks );
 346                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 347
 348                 foreach( $matches as $marker => $data ) {
 349                         list( $element, $content, $params, $tag ) = $data;
 350                         $tagName = strtolower( $element );
 351                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 352                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 353                                         array( $content, $params, $this ) );
 354                         } else {
 355                                 $output = $tag;
 356                         }
 357                         $this->mStripState->general->setPair( $marker, $output );
 358                 }
 359                 $text = $this->mStripState->unstripGeneral( $text );
 360
 361                 $text = Sanitizer::normalizeCharReferences( $text );
 362
 363                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 364                         $text = Parser::tidy($text);
 365                 } else {
 366                         # attempt to sanitize at least some nesting problems
 367                         # (bug #2702 and quite a few others)
 368                         $tidyregs = array(
 369                                 # ''Something [http://www.cool.com cool''] -->
 370                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 371                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 372                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 373                                 # fix up an anchor inside another anchor, only
 374                                 # at least for a single single nested link (bug 3695)
 375                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 376                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 377                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 378                                 # contains a div, so fix it up here; replace
 379                                 # div with escaped text
 380                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 381                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 382                                 # remove empty italic or bold tag pairs, some
 383                                 # introduced by rules above
 384                                 '/<([bi])><\/\\1>/' => '',
 385                         );
 386
 387                         $text = preg_replace(
 388                                 array_keys( $tidyregs ),
 389                                 array_values( $tidyregs ),
 390                                 $text );
 391                 }
 392
 393                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 394
 395                 # Information on include size limits, for the benefit of users who try to skirt them
 396                 if ( $this->mOptions->getEnableLimitReport() ) {
 397                         $max = $this->mOptions->getMaxIncludeSize();
 398                         $limitReport =
 399                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 400                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 401                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 402                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 403                         $text .= "\n<!-- \n$limitReport-->\n";
 404                 }
 405                 $this->mOutput->setText( $text );
 406                 $this->mRevisionId = $oldRevisionId;
 407                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 408                 wfProfileOut( $fname );
 409                 wfProfileOut( __METHOD__ );
 410
 411                 return $this->mOutput;
 412         }
 413
 414         /**
 415          * Recursive parser entry point that can be called from an extension tag
 416          * hook.
 417          */
 418         function recursiveTagParse( $text ) {
 419                 wfProfileIn( __METHOD__ );
 420                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 421                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 422                 $text = $this->internalParse( $text );
 423                 wfProfileOut( __METHOD__ );
 424                 return $text;
 425         }
 426
 427         /**
 428          * Expand templates and variables in the text, producing valid, static wikitext.
 429          * Also removes comments.
 430          */
 431         function preprocess( $text, $title, $options, $revid = null ) {
 432                 wfProfileIn( __METHOD__ );
 433                 $this->clearState();
 434                 $this->setOutputType( OT_PREPROCESS );
 435                 $this->mOptions = $options;
 436                 $this->mTitle = $title;
 437                 if( $revid !== null ) {
 438                         $this->mRevisionId = $revid;
 439                 }
 440                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 441                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 442                 $text = $this->replaceVariables( $text );
 443                 if ( $this->mOptions->getRemoveComments() ) {
 444                         $text = Sanitizer::removeHTMLcomments( $text );
 445                 }
 446                 $text = $this->mStripState->unstripBoth( $text );
 447                 wfProfileOut( __METHOD__ );
 448                 return $text;
 449         }
 450
 451         /**
 452          * Get a random string
 453          *
 454          * @private
 455          * @static
 456          */
 457         function getRandomString() {
 458                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 459         }
 460
 461         function &getTitle() { return $this->mTitle; }
 462         function getOptions() { return $this->mOptions; }
 463
 464         function getFunctionLang() {
 465                 global $wgLang, $wgContLang;
 466                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 467         }
 468
 469         /**
 470          * Replaces all occurrences of HTML-style comments and the given tags
 471          * in the text with a random marker and returns teh next text. The output
 472          * parameter $matches will be an associative array filled with data in
 473          * the form:
 474          *   'UNIQ-xxxxx' => array(
 475          *     'element',
 476          *     'tag content',
 477          *     array( 'param' => 'x' ),
 478          *     '<element param="x">tag content</element>' ) )
 479          *
 480          * @param $elements list of element names. Comments are always extracted.
 481          * @param $text Source text string.
 482          * @param $uniq_prefix
 483          *
 484          * @public
 485          * @static
 486          */
 487         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 488                 static $n = 1;
 489                 $stripped = '';
 490                 $matches = array();
 491
 492                 $taglist = implode( '|', $elements );
 493                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 494
 495                 while ( '' != $text ) {
 496                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 497                         $stripped .= $p[0];
 498                         if( count( $p ) < 5 ) {
 499                                 break;
 500                         }
 501                         if( count( $p ) > 5 ) {
 502                                 // comment
 503                                 $element    = $p[4];
 504                                 $attributes = '';
 505                                 $close      = '';
 506                                 $inside     = $p[5];
 507                         } else {
 508                                 // tag
 509                                 $element    = $p[1];
 510                                 $attributes = $p[2];
 511                                 $close      = $p[3];
 512                                 $inside     = $p[4];
 513                         }
 514
 515                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 516                         $stripped .= $marker;
 517
 518                         if ( $close === '/>' ) {
 519                                 // Empty element tag, <tag />
 520                                 $content = null;
 521                                 $text = $inside;
 522                                 $tail = null;
 523                         } else {
 524                                 if( $element == '!--' ) {
 525                                         $end = '/(-->)/';
 526                                 } else {
 527                                         $end = "/(<\\/$element\\s*>)/i";
 528                                 }
 529                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 530                                 $content = $q[0];
 531                                 if( count( $q ) < 3 ) {
 532                                         # No end tag -- let it run out to the end of the text.
 533                                         $tail = '';
 534                                         $text = '';
 535                                 } else {
 536                                         $tail = $q[1];
 537                                         $text = $q[2];
 538                                 }
 539                         }
 540
 541                         $matches[$marker] = array( $element,
 542                                 $content,
 543                                 Sanitizer::decodeTagAttributes( $attributes ),
 544                                 "<$element$attributes$close$content$tail" );
 545                 }
 546                 return $stripped;
 547         }
 548
 549         /**
 550          * Get a list of strippable XML-like elements
 551          */
 552         function getStripList() {
 553                 global $wgRawHtml;
 554                 $elements = $this->mStripList;
 555                 if( $wgRawHtml ) {
 556                         $elements[] = 'html';
 557                 }
 558                 if( $this->mOptions->getUseTeX() ) {
 559                         $elements[] = 'math';
 560                 }
 561                 return $elements;
 562         }
 563
 564         /**
 565          * @deprecated use replaceVariables
 566          */
 567         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 568                 return $text;
 569         }
 570
 571         /**
 572          * Restores pre, math, and other extensions removed by strip()
 573          *
 574          * always call unstripNoWiki() after this one
 575          * @private
 576          * @deprecated use $this->mStripState->unstrip()
 577          */
 578         function unstrip( $text, $state ) {
 579                 return $state->unstripGeneral( $text );
 580         }
 581
 582         /**
 583          * Always call this after unstrip() to preserve the order
 584          *
 585          * @private
 586          * @deprecated use $this->mStripState->unstrip()
 587          */
 588         function unstripNoWiki( $text, $state ) {
 589                 return $state->unstripNoWiki( $text );
 590         }
 591
 592         /**
 593          * @deprecated use $this->mStripState->unstripBoth()
 594          */
 595         function unstripForHTML( $text ) {
 596                 return $this->mStripState->unstripBoth( $text );
 597         }
 598
 599         /**
 600          * Add an item to the strip state
 601          * Returns the unique tag which must be inserted into the stripped text
 602          * The tag will be replaced with the original text in unstrip()
 603          *
 604          * @private
 605          */
 606         function insertStripItem( $text ) {
 607                 static $n = 0;
 608                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 609                 ++$n;
 610                 $this->mStripState->general->setPair( $rnd, $text );
 611                 return $rnd;
 612         }
 613
 614         /**
 615          * Interface with html tidy, used if $wgUseTidy = true.
 616          * If tidy isn't able to correct the markup, the original will be
 617          * returned in all its glory with a warning comment appended.
 618          *
 619          * Either the external tidy program or the in-process tidy extension
 620          * will be used depending on availability. Override the default
 621          * $wgTidyInternal setting to disable the internal if it's not working.
 622          *
 623          * @param string $text Hideous HTML input
 624          * @return string Corrected HTML output
 625          * @public
 626          * @static
 627          */
 628         function tidy( $text ) {
 629                 global $wgTidyInternal;
 630                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 631 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 632 '<head><title>test</title></head><body>'.$text.'</body></html>';
 633                 if( $wgTidyInternal ) {
 634                         $correctedtext = Parser::internalTidy( $wrappedtext );
 635                 } else {
 636                         $correctedtext = Parser::externalTidy( $wrappedtext );
 637                 }
 638                 if( is_null( $correctedtext ) ) {
 639                         wfDebug( "Tidy error detected!\n" );
 640                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 641                 }
 642                 return $correctedtext;
 643         }
 644
 645         /**
 646          * Spawn an external HTML tidy process and get corrected markup back from it.
 647          *
 648          * @private
 649          * @static
 650          */
 651         function externalTidy( $text ) {
 652                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 653                 $fname = 'Parser::externalTidy';
 654                 wfProfileIn( $fname );
 655
 656                 $cleansource = '';
 657                 $opts = ' -utf8';
 658
 659                 $descriptorspec = array(
 660                         0 => array('pipe', 'r'),
 661                         1 => array('pipe', 'w'),
 662                         2 => array('file', wfGetNull(), 'a')
 663                 );
 664                 $pipes = array();
 665                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 666                 if (is_resource($process)) {
 667                         // Theoretically, this style of communication could cause a deadlock
 668                         // here. If the stdout buffer fills up, then writes to stdin could
 669                         // block. This doesn't appear to happen with tidy, because tidy only
 670                         // writes to stdout after it's finished reading from stdin. Search
 671                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 672                         fwrite($pipes[0], $text);
 673                         fclose($pipes[0]);
 674                         while (!feof($pipes[1])) {
 675                                 $cleansource .= fgets($pipes[1], 1024);
 676                         }
 677                         fclose($pipes[1]);
 678                         proc_close($process);
 679                 }
 680
 681                 wfProfileOut( $fname );
 682
 683                 if( $cleansource == '' && $text != '') {
 684                         // Some kind of error happened, so we couldn't get the corrected text.
 685                         // Just give up; we'll use the source text and append a warning.
 686                         return null;
 687                 } else {
 688                         return $cleansource;
 689                 }
 690         }
 691
 692         /**
 693          * Use the HTML tidy PECL extension to use the tidy library in-process,
 694          * saving the overhead of spawning a new process.
 695          *
 696          * 'pear install tidy' should be able to compile the extension module.
 697          *
 698          * @private
 699          * @static
 700          */
 701         function internalTidy( $text ) {
 702                 global $wgTidyConf, $IP, $wgDebugTidy;
 703                 $fname = 'Parser::internalTidy';
 704                 wfProfileIn( $fname );
 705
 706                 $tidy = new tidy;
 707                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 708                 $tidy->cleanRepair();
 709                 if( $tidy->getStatus() == 2 ) {
 710                         // 2 is magic number for fatal error
 711                         // http://www.php.net/manual/en/function.tidy-get-status.php
 712                         $cleansource = null;
 713                 } else {
 714                         $cleansource = tidy_get_output( $tidy );
 715                 }
 716                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 717                         $cleansource .= "<!--\nTidy reports:\n" .
 718                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 719                                 "\n-->";
 720                 }
 721
 722                 wfProfileOut( $fname );
 723                 return $cleansource;
 724         }
 725
 726         /**
 727          * parse the wiki syntax used to render tables
 728          *
 729          * @private
 730          */
 731         function doTableStuff ( $text ) {
 732                 $fname = 'Parser::doTableStuff';
 733                 wfProfileIn( $fname );
 734
 735                 $lines = explode ( "\n" , $text );
 736                 $td_history = array (); // Is currently a td tag open?
 737                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 738                 $tr_history = array (); // Is currently a tr tag open?
 739                 $tr_attributes = array (); // history of tr attributes
 740                 $has_opened_tr = array(); // Did this table open a <tr> element?
 741                 $indent_level = 0; // indent level of the table
 742                 foreach ( $lines as $key => $line )
 743                 {
 744                         $line = trim ( $line );
 745
 746                         if( $line == '' ) { // empty line, go to next line
 747                                 continue;
 748                         }
 749                         $first_character = $line{0};
 750                         $matches = array();
 751
 752                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 753                                 // First check if we are starting a new table
 754                                 $indent_level = strlen( $matches[1] );
 755
 756                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 757                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 758
 759                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 760                                 array_push ( $td_history , false );
 761                                 array_push ( $last_tag_history , '' );
 762                                 array_push ( $tr_history , false );
 763                                 array_push ( $tr_attributes , '' );
 764                                 array_push ( $has_opened_tr , false );
 765                         } else if ( count ( $td_history ) == 0 ) {
 766                                 // Don't do any of the following
 767                                 continue;
 768                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 769                                 // We are ending a table
 770                                 $line = '</table>' . substr ( $line , 2 );
 771                                 $last_tag = array_pop ( $last_tag_history );
 772
 773                                 if ( !array_pop ( $has_opened_tr ) ) {
 774                                         $line = "<tr><td></td></tr>{$line}";
 775                                 }
 776
 777                                 if ( array_pop ( $tr_history ) ) {
 778                                         $line = "</tr>{$line}";
 779                                 }
 780
 781                                 if ( array_pop ( $td_history ) ) {
 782                                         $line = "</{$last_tag}>{$line}";
 783                                 }
 784                                 array_pop ( $tr_attributes );
 785                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 786                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 787                                 // Now we have a table row
 788                                 $line = preg_replace( '#^\|-+#', '', $line );
 789
 790                                 // Whats after the tag is now only attributes
 791                                 $attributes = $this->mStripState->unstripBoth( $line );
 792                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 793                                 array_pop ( $tr_attributes );
 794                                 array_push ( $tr_attributes , $attributes );
 795
 796                                 $line = '';
 797                                 $last_tag = array_pop ( $last_tag_history );
 798                                 array_pop ( $has_opened_tr );
 799                                 array_push ( $has_opened_tr , true );
 800
 801                                 if ( array_pop ( $tr_history ) ) {
 802                                         $line = '</tr>';
 803                                 }
 804
 805                                 if ( array_pop ( $td_history ) ) {
 806                                         $line = "</{$last_tag}>{$line}";
 807                                 }
 808
 809                                 $lines[$key] = $line;
 810                                 array_push ( $tr_history , false );
 811                                 array_push ( $td_history , false );
 812                                 array_push ( $last_tag_history , '' );
 813                         }
 814                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 815                                 // This might be cell elements, td, th or captions
 816                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 817                                         $first_character = '+';
 818                                         $line = substr ( $line , 1 );
 819                                 }
 820
 821                                 $line = substr ( $line , 1 );
 822
 823                                 if ( $first_character == '!' ) {
 824                                         $line = str_replace ( '!!' , '||' , $line );
 825                                 }
 826
 827                                 // Split up multiple cells on the same line.
 828                                 // FIXME : This can result in improper nesting of tags processed
 829                                 // by earlier parser steps, but should avoid splitting up eg
 830                                 // attribute values containing literal "||".
 831                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 832
 833                                 $lines[$key] = '';
 834
 835                                 // Loop through each table cell
 836                                 foreach ( $cells as $cell )
 837                                 {
 838                                         $previous = '';
 839                                         if ( $first_character != '+' )
 840                                         {
 841                                                 $tr_after = array_pop ( $tr_attributes );
 842                                                 if ( !array_pop ( $tr_history ) ) {
 843                                                         $previous = "<tr{$tr_after}>\n";
 844                                                 }
 845                                                 array_push ( $tr_history , true );
 846                                                 array_push ( $tr_attributes , '' );
 847                                                 array_pop ( $has_opened_tr );
 848                                                 array_push ( $has_opened_tr , true );
 849                                         }
 850
 851                                         $last_tag = array_pop ( $last_tag_history );
 852
 853                                         if ( array_pop ( $td_history ) ) {
 854                                                 $previous = "</{$last_tag}>{$previous}";
 855                                         }
 856
 857                                         if ( $first_character == '|' ) {
 858                                                 $last_tag = 'td';
 859                                         } else if ( $first_character == '!' ) {
 860                                                 $last_tag = 'th';
 861                                         } else if ( $first_character == '+' ) {
 862                                                 $last_tag = 'caption';
 863                                         } else {
 864                                                 $last_tag = '';
 865                                         }
 866
 867                                         array_push ( $last_tag_history , $last_tag );
 868
 869                                         // A cell could contain both parameters and data
 870                                         $cell_data = explode ( '|' , $cell , 2 );
 871
 872                                         // Bug 553: Note that a '|' inside an invalid link should not
 873                                         // be mistaken as delimiting cell parameters
 874                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 875                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 876                                         } else if ( count ( $cell_data ) == 1 )
 877                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 878                                         else {
 879                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 880                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 881                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 882                                         }
 883
 884                                         $lines[$key] .= $cell;
 885                                         array_push ( $td_history , true );
 886                                 }
 887                         }
 888                 }
 889
 890                 // Closing open td, tr && table
 891                 while ( count ( $td_history ) > 0 )
 892                 {
 893                         if ( array_pop ( $td_history ) ) {
 894                                 $lines[] = '</td>' ;
 895                         }
 896                         if ( array_pop ( $tr_history ) ) {
 897                                 $lines[] = '</tr>' ;
 898                         }
 899                         if ( !array_pop ( $has_opened_tr ) ) {
 900                                 $lines[] = "<tr><td></td></tr>" ;
 901                         }
 902
 903                         $lines[] = '</table>' ;
 904                 }
 905
 906                 $output = implode ( "\n" , $lines ) ;
 907
 908                 // special case: don't return empty table
 909                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 910                         $output = '';
 911                 }
 912
 913                 wfProfileOut( $fname );
 914
 915                 return $output;
 916         }
 917
 918         /**
 919          * Helper function for parse() that transforms wiki markup into
 920          * HTML. Only called for $mOutputType == OT_HTML.
 921          *
 922          * @private
 923          */
 924         function internalParse( $text ) {
 925                 $isMain = true;
 926                 $fname = 'Parser::internalParse';
 927                 wfProfileIn( $fname );
 928
 929                 # Hook to suspend the parser in this state
 930                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 931                         wfProfileOut( $fname );
 932                         return $text ;
 933                 }
 934
 935                 $text = $this->replaceVariables( $text );
 936                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 937                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 938
 939                 // Tables need to come after variable replacement for things to work
 940                 // properly; putting them before other transformations should keep
 941                 // exciting things like link expansions from showing up in surprising
 942                 // places.
 943                 $text = $this->doTableStuff( $text );
 944
 945                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 946
 947                 $text = $this->stripToc( $text );
 948                 $this->stripNoGallery( $text );
 949                 $text = $this->doHeadings( $text );
 950                 if($this->mOptions->getUseDynamicDates()) {
 951                         $df =& DateFormatter::getInstance();
 952                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 953                 }
 954                 $text = $this->doAllQuotes( $text );
 955                 $text = $this->replaceInternalLinks( $text );
 956                 $text = $this->replaceExternalLinks( $text );
 957
 958                 # replaceInternalLinks may sometimes leave behind
 959                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 960                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 961
 962                 $text = $this->doMagicLinks( $text );
 963                 $text = $this->formatHeadings( $text, $isMain );
 964
 965                 wfProfileOut( $fname );
 966                 return $text;
 967         }
 968
 969         /**
 970          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 971          * magic external links.
 972          *
 973          * @private
 974          */
 975         function doMagicLinks( $text ) {
 976                 wfProfileIn( __METHOD__ );
 977                 $text = preg_replace_callback(
 978                         '!(?:                           # Start cases
 979                             <a.*?</a> |                 # Skip link text
 980                             <.*?> |                     # Skip stuff inside HTML elements
 981                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 982                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 983                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 984                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 985                                      [0-9Xx]                 # check digit
 986                                    \b)
 987                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 988                 wfProfileOut( __METHOD__ );
 989                 return $text;
 990         }
 991
 992         function magicLinkCallback( $m ) {
 993                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 994                         # Skip HTML element
 995                         return $m[0];
 996                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 997                         $isbn = $m[2];
 998                         $num = strtr( $isbn, array(
 999                                 '-' => '',
1000                                 ' ' => '',
1001                                 'x' => 'X',
1002                         ));
1003                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1004                         $text = '<a href="' .
1005                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1006                                 "\" class=\"internal\">ISBN $isbn</a>";
1007                 } else {
1008                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1009                                 $keyword = 'RFC';
1010                                 $urlmsg = 'rfcurl';
1011                                 $id = $m[1];
1012                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1013                                 $keyword = 'PMID';
1014                                 $urlmsg = 'pubmedurl';
1015                                 $id = $m[1];
1016                         } else {
1017                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1018                                         substr($m[0], 0, 20 ) . '"' );
1019                         }
1020
1021                         $url = wfMsg( $urlmsg, $id);
1022                         $sk = $this->mOptions->getSkin();
1023                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1024                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1025                 }
1026                 return $text;
1027         }
1028
1029         /**
1030          * Parse headers and return html
1031          *
1032          * @private
1033          */
1034         function doHeadings( $text ) {
1035                 $fname = 'Parser::doHeadings';
1036                 wfProfileIn( $fname );
1037                 for ( $i = 6; $i >= 1; --$i ) {
1038                         $h = str_repeat( '=', $i );
1039                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1040                           "<h$i>\\1</h$i>", $text );
1041                 }
1042                 wfProfileOut( $fname );
1043                 return $text;
1044         }
1045
1046         /**
1047          * Replace single quotes with HTML markup
1048          * @private
1049          * @return string the altered text
1050          */
1051         function doAllQuotes( $text ) {
1052                 $fname = 'Parser::doAllQuotes';
1053                 wfProfileIn( $fname );
1054                 $outtext = '';
1055                 $lines = explode( "\n", $text );
1056                 foreach ( $lines as $line ) {
1057                         $outtext .= $this->doQuotes ( $line ) . "\n";
1058                 }
1059                 $outtext = substr($outtext, 0,-1);
1060                 wfProfileOut( $fname );
1061                 return $outtext;
1062         }
1063
1064         /**
1065          * Helper function for doAllQuotes()
1066          */
1067         public function doQuotes( $text ) {
1068                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1069                 if ( count( $arr ) == 1 )
1070                         return $text;
1071                 else
1072                 {
1073                         # First, do some preliminary work. This may shift some apostrophes from
1074                         # being mark-up to being text. It also counts the number of occurrences
1075                         # of bold and italics mark-ups.
1076                         $i = 0;
1077                         $numbold = 0;
1078                         $numitalics = 0;
1079                         foreach ( $arr as $r )
1080                         {
1081                                 if ( ( $i % 2 ) == 1 )
1082                                 {
1083                                         # If there are ever four apostrophes, assume the first is supposed to
1084                                         # be text, and the remaining three constitute mark-up for bold text.
1085                                         if ( strlen( $arr[$i] ) == 4 )
1086                                         {
1087                                                 $arr[$i-1] .= "'";
1088                                                 $arr[$i] = "'''";
1089                                         }
1090                                         # If there are more than 5 apostrophes in a row, assume they're all
1091                                         # text except for the last 5.
1092                                         else if ( strlen( $arr[$i] ) > 5 )
1093                                         {
1094                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1095                                                 $arr[$i] = "'''''";
1096                                         }
1097                                         # Count the number of occurrences of bold and italics mark-ups.
1098                                         # We are not counting sequences of five apostrophes.
1099                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1100                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1101                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1102                                 }
1103                                 $i++;
1104                         }
1105
1106                         # If there is an odd number of both bold and italics, it is likely
1107                         # that one of the bold ones was meant to be an apostrophe followed
1108                         # by italics. Which one we cannot know for certain, but it is more
1109                         # likely to be one that has a single-letter word before it.
1110                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1111                         {
1112                                 $i = 0;
1113                                 $firstsingleletterword = -1;
1114                                 $firstmultiletterword = -1;
1115                                 $firstspace = -1;
1116                                 foreach ( $arr as $r )
1117                                 {
1118                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1119                                         {
1120                                                 $x1 = substr ($arr[$i-1], -1);
1121                                                 $x2 = substr ($arr[$i-1], -2, 1);
1122                                                 if ($x1 == ' ') {
1123                                                         if ($firstspace == -1) $firstspace = $i;
1124                                                 } else if ($x2 == ' ') {
1125                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1126                                                 } else {
1127                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1128                                                 }
1129                                         }
1130                                         $i++;
1131                                 }
1132
1133                                 # If there is a single-letter word, use it!
1134                                 if ($firstsingleletterword > -1)
1135                                 {
1136                                         $arr [ $firstsingleletterword ] = "''";
1137                                         $arr [ $firstsingleletterword-1 ] .= "'";
1138                                 }
1139                                 # If not, but there's a multi-letter word, use that one.
1140                                 else if ($firstmultiletterword > -1)
1141                                 {
1142                                         $arr [ $firstmultiletterword ] = "''";
1143                                         $arr [ $firstmultiletterword-1 ] .= "'";
1144                                 }
1145                                 # ... otherwise use the first one that has neither.
1146                                 # (notice that it is possible for all three to be -1 if, for example,
1147                                 # there is only one pentuple-apostrophe in the line)
1148                                 else if ($firstspace > -1)
1149                                 {
1150                                         $arr [ $firstspace ] = "''";
1151                                         $arr [ $firstspace-1 ] .= "'";
1152                                 }
1153                         }
1154
1155                         # Now let's actually convert our apostrophic mush to HTML!
1156                         $output = '';
1157                         $buffer = '';
1158                         $state = '';
1159                         $i = 0;
1160                         foreach ($arr as $r)
1161                         {
1162                                 if (($i % 2) == 0)
1163                                 {
1164                                         if ($state == 'both')
1165                                                 $buffer .= $r;
1166                                         else
1167                                                 $output .= $r;
1168                                 }
1169                                 else
1170                                 {
1171                                         if (strlen ($r) == 2)
1172                                         {
1173                                                 if ($state == 'i')
1174                                                 { $output .= '</i>'; $state = ''; }
1175                                                 else if ($state == 'bi')
1176                                                 { $output .= '</i>'; $state = 'b'; }
1177                                                 else if ($state == 'ib')
1178                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1179                                                 else if ($state == 'both')
1180                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1181                                                 else # $state can be 'b' or ''
1182                                                 { $output .= '<i>'; $state .= 'i'; }
1183                                         }
1184                                         else if (strlen ($r) == 3)
1185                                         {
1186                                                 if ($state == 'b')
1187                                                 { $output .= '</b>'; $state = ''; }
1188                                                 else if ($state == 'bi')
1189                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1190                                                 else if ($state == 'ib')
1191                                                 { $output .= '</b>'; $state = 'i'; }
1192                                                 else if ($state == 'both')
1193                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1194                                                 else # $state can be 'i' or ''
1195                                                 { $output .= '<b>'; $state .= 'b'; }
1196                                         }
1197                                         else if (strlen ($r) == 5)
1198                                         {
1199                                                 if ($state == 'b')
1200                                                 { $output .= '</b><i>'; $state = 'i'; }
1201                                                 else if ($state == 'i')
1202                                                 { $output .= '</i><b>'; $state = 'b'; }
1203                                                 else if ($state == 'bi')
1204                                                 { $output .= '</i></b>'; $state = ''; }
1205                                                 else if ($state == 'ib')
1206                                                 { $output .= '</b></i>'; $state = ''; }
1207                                                 else if ($state == 'both')
1208                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1209                                                 else # ($state == '')
1210                                                 { $buffer = ''; $state = 'both'; }
1211                                         }
1212                                 }
1213                                 $i++;
1214                         }
1215                         # Now close all remaining tags.  Notice that the order is important.
1216                         if ($state == 'b' || $state == 'ib')
1217                                 $output .= '</b>';
1218                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1219                                 $output .= '</i>';
1220                         if ($state == 'bi')
1221                                 $output .= '</b>';
1222                         # There might be lonely ''''', so make sure we have a buffer
1223                         if ($state == 'both' && $buffer)
1224                                 $output .= '<b><i>'.$buffer.'</i></b>';
1225                         return $output;
1226                 }
1227         }
1228
1229         /**
1230          * Replace external links
1231          *
1232          * Note: this is all very hackish and the order of execution matters a lot.
1233          * Make sure to run maintenance/parserTests.php if you change this code.
1234          *
1235          * @private
1236          */
1237         function replaceExternalLinks( $text ) {
1238                 global $wgContLang;
1239                 $fname = 'Parser::replaceExternalLinks';
1240                 wfProfileIn( $fname );
1241
1242                 $sk = $this->mOptions->getSkin();
1243
1244                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1245
1246                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1247
1248                 $i = 0;
1249                 while ( $i<count( $bits ) ) {
1250                         $url = $bits[$i++];
1251                         $protocol = $bits[$i++];
1252                         $text = $bits[$i++];
1253                         $trail = $bits[$i++];
1254
1255                         # The characters '<' and '>' (which were escaped by
1256                         # removeHTMLtags()) should not be included in
1257                         # URLs, per RFC 2396.
1258                         $m2 = array();
1259                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1260                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1261                                 $url = substr($url, 0, $m2[0][1]);
1262                         }
1263
1264                         # If the link text is an image URL, replace it with an <img> tag
1265                         # This happened by accident in the original parser, but some people used it extensively
1266                         $img = $this->maybeMakeExternalImage( $text );
1267                         if ( $img !== false ) {
1268                                 $text = $img;
1269                         }
1270
1271                         $dtrail = '';
1272
1273                         # Set linktype for CSS - if URL==text, link is essentially free
1274                         $linktype = ($text == $url) ? 'free' : 'text';
1275
1276                         # No link text, e.g. [http://domain.tld/some.link]
1277                         if ( $text == '' ) {
1278                                 # Autonumber if allowed. See bug #5918
1279                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1280                                         $text = '[' . ++$this->mAutonumber . ']';
1281                                         $linktype = 'autonumber';
1282                                 } else {
1283                                         # Otherwise just use the URL
1284                                         $text = htmlspecialchars( $url );
1285                                         $linktype = 'free';
1286                                 }
1287                         } else {
1288                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1289                                 # Check for trail
1290                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1291                         }
1292
1293                         $text = $wgContLang->markNoConversion($text);
1294
1295                         $url = Sanitizer::cleanUrl( $url );
1296
1297                         # Process the trail (i.e. everything after this link up until start of the next link),
1298                         # replacing any non-bracketed links
1299                         $trail = $this->replaceFreeExternalLinks( $trail );
1300
1301                         # Use the encoded URL
1302                         # This means that users can paste URLs directly into the text
1303                         # Funny characters like &ouml; aren't valid in URLs anyway
1304                         # This was changed in August 2004
1305                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1306
1307                         # Register link in the output object.
1308                         # Replace unnecessary URL escape codes with the referenced character
1309                         # This prevents spammers from hiding links from the filters
1310                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1311                         $this->mOutput->addExternalLink( $pasteurized );
1312                 }
1313
1314                 wfProfileOut( $fname );
1315                 return $s;
1316         }
1317
1318         /**
1319          * Replace anything that looks like a URL with a link
1320          * @private
1321          */
1322         function replaceFreeExternalLinks( $text ) {
1323                 global $wgContLang;
1324                 $fname = 'Parser::replaceFreeExternalLinks';
1325                 wfProfileIn( $fname );
1326
1327                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1328                 $s = array_shift( $bits );
1329                 $i = 0;
1330
1331                 $sk = $this->mOptions->getSkin();
1332
1333                 while ( $i < count( $bits ) ){
1334                         $protocol = $bits[$i++];
1335                         $remainder = $bits[$i++];
1336
1337                         $m = array();
1338                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1339                                 # Found some characters after the protocol that look promising
1340                                 $url = $protocol . $m[1];
1341                                 $trail = $m[2];
1342
1343                                 # special case: handle urls as url args:
1344                                 # http://www.example.com/foo?=http://www.example.com/bar
1345                                 if(strlen($trail) == 0 &&
1346                                         isset($bits[$i]) &&
1347                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1348                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1349                                 {
1350                                         # add protocol, arg
1351                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1352                                         $i += 2;
1353                                         $trail = $m[2];
1354                                 }
1355
1356                                 # The characters '<' and '>' (which were escaped by
1357                                 # removeHTMLtags()) should not be included in
1358                                 # URLs, per RFC 2396.
1359                                 $m2 = array();
1360                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1361                                         $trail = substr($url, $m2[0][1]) . $trail;
1362                                         $url = substr($url, 0, $m2[0][1]);
1363                                 }
1364
1365                                 # Move trailing punctuation to $trail
1366                                 $sep = ',;\.:!?';
1367                                 # If there is no left bracket, then consider right brackets fair game too
1368                                 if ( strpos( $url, '(' ) === false ) {
1369                                         $sep .= ')';
1370                                 }
1371
1372                                 $numSepChars = strspn( strrev( $url ), $sep );
1373                                 if ( $numSepChars ) {
1374                                         $trail = substr( $url, -$numSepChars ) . $trail;
1375                                         $url = substr( $url, 0, -$numSepChars );
1376                                 }
1377
1378                                 $url = Sanitizer::cleanUrl( $url );
1379
1380                                 # Is this an external image?
1381                                 $text = $this->maybeMakeExternalImage( $url );
1382                                 if ( $text === false ) {
1383                                         # Not an image, make a link
1384                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1385                                         # Register it in the output object...
1386                                         # Replace unnecessary URL escape codes with their equivalent characters
1387                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1388                                         $this->mOutput->addExternalLink( $pasteurized );
1389                                 }
1390                                 $s .= $text . $trail;
1391                         } else {
1392                                 $s .= $protocol . $remainder;
1393                         }
1394                 }
1395                 wfProfileOut( $fname );
1396                 return $s;
1397         }
1398
1399         /**
1400          * Replace unusual URL escape codes with their equivalent characters
1401          * @param string
1402          * @return string
1403          * @static
1404          * @todo  This can merge genuinely required bits in the path or query string,
1405          *        breaking legit URLs. A proper fix would treat the various parts of
1406          *        the URL differently; as a workaround, just use the output for
1407          *        statistical records, not for actual linking/output.
1408          */
1409         static function replaceUnusualEscapes( $url ) {
1410                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1411                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1412         }
1413
1414         /**
1415          * Callback function used in replaceUnusualEscapes().
1416          * Replaces unusual URL escape codes with their equivalent character
1417          * @static
1418          * @private
1419          */
1420         private static function replaceUnusualEscapesCallback( $matches ) {
1421                 $char = urldecode( $matches[0] );
1422                 $ord = ord( $char );
1423                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1424                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1425                         // No, shouldn't be escaped
1426                         return $char;
1427                 } else {
1428                         // Yes, leave it escaped
1429                         return $matches[0];
1430                 }
1431         }
1432
1433         /**
1434          * make an image if it's allowed, either through the global
1435          * option or through the exception
1436          * @private
1437          */
1438         function maybeMakeExternalImage( $url ) {
1439                 $sk = $this->mOptions->getSkin();
1440                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1441                 $imagesexception = !empty($imagesfrom);
1442                 $text = false;
1443                 if ( $this->mOptions->getAllowExternalImages()
1444                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1445                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1446                                 # Image found
1447                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1448                         }
1449                 }
1450                 return $text;
1451         }
1452
1453         /**
1454          * Process [[ ]] wikilinks
1455          *
1456          * @private
1457          */
1458         function replaceInternalLinks( $s ) {
1459                 global $wgContLang;
1460                 static $fname = 'Parser::replaceInternalLinks' ;
1461
1462                 wfProfileIn( $fname );
1463
1464                 wfProfileIn( $fname.'-setup' );
1465                 static $tc = FALSE;
1466                 # the % is needed to support urlencoded titles as well
1467                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1468
1469                 $sk = $this->mOptions->getSkin();
1470
1471                 #split the entire text string on occurences of [[
1472                 $a = explode( '[[', ' ' . $s );
1473                 #get the first element (all text up to first [[), and remove the space we added
1474                 $s = array_shift( $a );
1475                 $s = substr( $s, 1 );
1476
1477                 # Match a link having the form [[namespace:link|alternate]]trail
1478                 static $e1 = FALSE;
1479                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1480                 # Match cases where there is no "]]", which might still be images
1481                 static $e1_img = FALSE;
1482                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1483
1484                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1485                 $e2 = null;
1486                 if ( $useLinkPrefixExtension ) {
1487                         # Match the end of a line for a word that's not followed by whitespace,
1488                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1489                         $e2 = wfMsgForContent( 'linkprefix' );
1490                 }
1491
1492                 if( is_null( $this->mTitle ) ) {
1493                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1494                 }
1495                 $nottalk = !$this->mTitle->isTalkPage();
1496
1497                 if ( $useLinkPrefixExtension ) {
1498                         $m = array();
1499                         if ( preg_match( $e2, $s, $m ) ) {
1500                                 $first_prefix = $m[2];
1501                         } else {
1502                                 $first_prefix = false;
1503                         }
1504                 } else {
1505                         $prefix = '';
1506                 }
1507
1508                 if($wgContLang->hasVariants()) {
1509                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1510                 } else {
1511                         $selflink = array($this->mTitle->getPrefixedText());
1512                 }
1513                 $useSubpages = $this->areSubpagesAllowed();
1514                 wfProfileOut( $fname.'-setup' );
1515
1516                 # Loop for each link
1517                 for ($k = 0; isset( $a[$k] ); $k++) {
1518                         $line = $a[$k];
1519                         if ( $useLinkPrefixExtension ) {
1520                                 wfProfileIn( $fname.'-prefixhandling' );
1521                                 if ( preg_match( $e2, $s, $m ) ) {
1522                                         $prefix = $m[2];
1523                                         $s = $m[1];
1524                                 } else {
1525                                         $prefix='';
1526                                 }
1527                                 # first link
1528                                 if($first_prefix) {
1529                                         $prefix = $first_prefix;
1530                                         $first_prefix = false;
1531                                 }
1532                                 wfProfileOut( $fname.'-prefixhandling' );
1533                         }
1534
1535                         $might_be_img = false;
1536
1537                         wfProfileIn( "$fname-e1" );
1538                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1539                                 $text = $m[2];
1540                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1541                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1542                                 # the real problem is with the $e1 regex
1543                                 # See bug 1300.
1544                                 #
1545                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1546                                 # and no image is in sight. See bug 2095.
1547                                 #
1548                                 if( $text !== '' &&
1549                                         substr( $m[3], 0, 1 ) === ']' &&
1550                                         strpos($text, '[') !== false
1551                                 )
1552                                 {
1553                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1554                                         $m[3] = substr( $m[3], 1 );
1555                                 }
1556                                 # fix up urlencoded title texts
1557                                 if( strpos( $m[1], '%' ) !== false ) {
1558                                         # Should anchors '#' also be rejected?
1559                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1560                                 }
1561                                 $trail = $m[3];
1562                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1563                                 $might_be_img = true;
1564                                 $text = $m[2];
1565                                 if ( strpos( $m[1], '%' ) !== false ) {
1566                                         $m[1] = urldecode($m[1]);
1567                                 }
1568                                 $trail = "";
1569                         } else { # Invalid form; output directly
1570                                 $s .= $prefix . '[[' . $line ;
1571                                 wfProfileOut( "$fname-e1" );
1572                                 continue;
1573                         }
1574                         wfProfileOut( "$fname-e1" );
1575                         wfProfileIn( "$fname-misc" );
1576
1577                         # Don't allow internal links to pages containing
1578                         # PROTO: where PROTO is a valid URL protocol; these
1579                         # should be external links.
1580                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1581                                 $s .= $prefix . '[[' . $line ;
1582                                 continue;
1583                         }
1584
1585                         # Make subpage if necessary
1586                         if( $useSubpages ) {
1587                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1588                         } else {
1589                                 $link = $m[1];
1590                         }
1591
1592                         $noforce = (substr($m[1], 0, 1) != ':');
1593                         if (!$noforce) {
1594                                 # Strip off leading ':'
1595                                 $link = substr($link, 1);
1596                         }
1597
1598                         wfProfileOut( "$fname-misc" );
1599                         wfProfileIn( "$fname-title" );
1600                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1601                         if( !$nt ) {
1602                                 $s .= $prefix . '[[' . $line;
1603                                 wfProfileOut( "$fname-title" );
1604                                 continue;
1605                         }
1606
1607                         $ns = $nt->getNamespace();
1608                         $iw = $nt->getInterWiki();
1609                         wfProfileOut( "$fname-title" );
1610
1611                         if ($might_be_img) { # if this is actually an invalid link
1612                                 wfProfileIn( "$fname-might_be_img" );
1613                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1614                                         $found = false;
1615                                         while (isset ($a[$k+1]) ) {
1616                                                 #look at the next 'line' to see if we can close it there
1617                                                 $spliced = array_splice( $a, $k + 1, 1 );
1618                                                 $next_line = array_shift( $spliced );
1619                                                 $m = explode( ']]', $next_line, 3 );
1620                                                 if ( count( $m ) == 3 ) {
1621                                                         # the first ]] closes the inner link, the second the image
1622                                                         $found = true;
1623                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1624                                                         $trail = $m[2];
1625                                                         break;
1626                                                 } elseif ( count( $m ) == 2 ) {
1627                                                         #if there's exactly one ]] that's fine, we'll keep looking
1628                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1629                                                 } else {
1630                                                         #if $next_line is invalid too, we need look no further
1631                                                         $text .= '[[' . $next_line;
1632                                                         break;
1633                                                 }
1634                                         }
1635                                         if ( !$found ) {
1636                                                 # we couldn't find the end of this imageLink, so output it raw
1637                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1638                                                 $text = $this->replaceInternalLinks($text);
1639                                                 $s .= "{$prefix}[[$link|$text";
1640                                                 # note: no $trail, because without an end, there *is* no trail
1641                                                 wfProfileOut( "$fname-might_be_img" );
1642                                                 continue;
1643                                         }
1644                                 } else { #it's not an image, so output it raw
1645                                         $s .= "{$prefix}[[$link|$text";
1646                                         # note: no $trail, because without an end, there *is* no trail
1647                                         wfProfileOut( "$fname-might_be_img" );
1648                                         continue;
1649                                 }
1650                                 wfProfileOut( "$fname-might_be_img" );
1651                         }
1652
1653                         $wasblank = ( '' == $text );
1654                         if( $wasblank ) $text = $link;
1655
1656                         # Link not escaped by : , create the various objects
1657                         if( $noforce ) {
1658
1659                                 # Interwikis
1660                                 wfProfileIn( "$fname-interwiki" );
1661                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1662                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1663                                         $s = rtrim($s . $prefix);
1664                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1665                                         wfProfileOut( "$fname-interwiki" );
1666                                         continue;
1667                                 }
1668                                 wfProfileOut( "$fname-interwiki" );
1669
1670                                 if ( $ns == NS_IMAGE ) {
1671                                         wfProfileIn( "$fname-image" );
1672                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1673                                                 # recursively parse links inside the image caption
1674                                                 # actually, this will parse them in any other parameters, too,
1675                                                 # but it might be hard to fix that, and it doesn't matter ATM
1676                                                 $text = $this->replaceExternalLinks($text);
1677                                                 $text = $this->replaceInternalLinks($text);
1678
1679                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1680                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1681                                                 $this->mOutput->addImage( $nt->getDBkey() );
1682
1683                                                 wfProfileOut( "$fname-image" );
1684                                                 continue;
1685                                         } else {
1686                                                 # We still need to record the image's presence on the page
1687                                                 $this->mOutput->addImage( $nt->getDBkey() );
1688                                         }
1689                                         wfProfileOut( "$fname-image" );
1690
1691                                 }
1692
1693                                 if ( $ns == NS_CATEGORY ) {
1694                                         wfProfileIn( "$fname-category" );
1695                                         $s = rtrim($s . "\n"); # bug 87
1696
1697                                         if ( $wasblank ) {
1698                                                 $sortkey = $this->getDefaultSort();
1699                                         } else {
1700                                                 $sortkey = $text;
1701                                         }
1702                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1703                                         $sortkey = str_replace( "\n", '', $sortkey );
1704                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1705                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1706
1707                                         /**
1708                                          * Strip the whitespace Category links produce, see bug 87
1709                                          * @todo We might want to use trim($tmp, "\n") here.
1710                                          */
1711                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1712
1713                                         wfProfileOut( "$fname-category" );
1714                                         continue;
1715                                 }
1716                         }
1717
1718                         # Self-link checking
1719                         if( $nt->getFragment() === '' ) {
1720                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1721                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1722                                         continue;
1723                                 }
1724                         }
1725
1726                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1727                         if( $ns == NS_MEDIA ) {
1728                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1729                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1730                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1731                                 $this->mOutput->addImage( $nt->getDBkey() );
1732                                 continue;
1733                         } elseif( $ns == NS_SPECIAL ) {
1734                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1735                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1736                                 } else {
1737                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1738                                 }
1739                                 continue;
1740                         } elseif( $ns == NS_IMAGE ) {
1741                                 $img = wfFindFile( $nt );
1742                                 if( $img ) {
1743                                         // Force a blue link if the file exists; may be a remote
1744                                         // upload on the shared repository, and we want to see its
1745                                         // auto-generated page.
1746                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1747                                         $this->mOutput->addLink( $nt );
1748                                         continue;
1749                                 }
1750                         }
1751                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1752                 }
1753                 wfProfileOut( $fname );
1754                 return $s;
1755         }
1756
1757         /**
1758          * Make a link placeholder. The text returned can be later resolved to a real link with
1759          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1760          * parsing of interwiki links, and secondly to allow all existence checks and
1761          * article length checks (for stub links) to be bundled into a single query.
1762          *
1763          */
1764         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1765                 wfProfileIn( __METHOD__ );
1766                 if ( ! is_object($nt) ) {
1767                         # Fail gracefully
1768                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1769                 } else {
1770                         # Separate the link trail from the rest of the link
1771                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1772
1773                         if ( $nt->isExternal() ) {
1774                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1775                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1776                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1777                         } else {
1778                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1779                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1780                                 $this->mLinkHolders['queries'][] = $query;
1781                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1782                                 $this->mLinkHolders['titles'][] = $nt;
1783
1784                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1785                         }
1786                 }
1787                 wfProfileOut( __METHOD__ );
1788                 return $retVal;
1789         }
1790
1791         /**
1792          * Render a forced-blue link inline; protect against double expansion of
1793          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1794          * Since this little disaster has to split off the trail text to avoid
1795          * breaking URLs in the following text without breaking trails on the
1796          * wiki links, it's been made into a horrible function.
1797          *
1798          * @param Title $nt
1799          * @param string $text
1800          * @param string $query
1801          * @param string $trail
1802          * @param string $prefix
1803          * @return string HTML-wikitext mix oh yuck
1804          */
1805         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1806                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1807                 $sk = $this->mOptions->getSkin();
1808                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1809                 return $this->armorLinks( $link ) . $trail;
1810         }
1811
1812         /**
1813          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1814          * going to go through further parsing steps before inline URL expansion.
1815          *
1816          * In particular this is important when using action=render, which causes
1817          * full URLs to be included.
1818          *
1819          * Oh man I hate our multi-layer parser!
1820          *
1821          * @param string more-or-less HTML
1822          * @return string less-or-more HTML with NOPARSE bits
1823          */
1824         function armorLinks( $text ) {
1825                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1826                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1827         }
1828
1829         /**
1830          * Return true if subpage links should be expanded on this page.
1831          * @return bool
1832          */
1833         function areSubpagesAllowed() {
1834                 # Some namespaces don't allow subpages
1835                 global $wgNamespacesWithSubpages;
1836                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1837         }
1838
1839         /**
1840          * Handle link to subpage if necessary
1841          * @param string $target the source of the link
1842          * @param string &$text the link text, modified as necessary
1843          * @return string the full name of the link
1844          * @private
1845          */
1846         function maybeDoSubpageLink($target, &$text) {
1847                 # Valid link forms:
1848                 # Foobar -- normal
1849                 # :Foobar -- override special treatment of prefix (images, language links)
1850                 # /Foobar -- convert to CurrentPage/Foobar
1851                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1852                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1853                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1854
1855                 $fname = 'Parser::maybeDoSubpageLink';
1856                 wfProfileIn( $fname );
1857                 $ret = $target; # default return value is no change
1858
1859                 # Some namespaces don't allow subpages,
1860                 # so only perform processing if subpages are allowed
1861                 if( $this->areSubpagesAllowed() ) {
1862                         $hash = strpos( $target, '#' );
1863                         if( $hash !== false ) {
1864                                 $suffix = substr( $target, $hash );
1865                                 $target = substr( $target, 0, $hash );
1866                         } else {
1867                                 $suffix = '';
1868                         }
1869                         # bug 7425
1870                         $target = trim( $target );
1871                         # Look at the first character
1872                         if( $target != '' && $target{0} == '/' ) {
1873                                 # / at end means we don't want the slash to be shown
1874                                 $m = array();
1875                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1876                                 if( $trailingSlashes ) {
1877                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1878                                 } else {
1879                                         $noslash = substr( $target, 1 );
1880                                 }
1881
1882                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1883                                 if( '' === $text ) {
1884                                         $text = $target . $suffix;
1885                                 } # this might be changed for ugliness reasons
1886                         } else {
1887                                 # check for .. subpage backlinks
1888                                 $dotdotcount = 0;
1889                                 $nodotdot = $target;
1890                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1891                                         ++$dotdotcount;
1892                                         $nodotdot = substr( $nodotdot, 3 );
1893                                 }
1894                                 if($dotdotcount > 0) {
1895                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1896                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1897                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1898                                                 # / at the end means don't show full path
1899                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1900                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1901                                                         if( '' === $text ) {
1902                                                                 $text = $nodotdot . $suffix;
1903                                                         }
1904                                                 }
1905                                                 $nodotdot = trim( $nodotdot );
1906                                                 if( $nodotdot != '' ) {
1907                                                         $ret .= '/' . $nodotdot;
1908                                                 }
1909                                                 $ret .= $suffix;
1910                                         }
1911                                 }
1912                         }
1913                 }
1914
1915                 wfProfileOut( $fname );
1916                 return $ret;
1917         }
1918
1919         /**#@+
1920          * Used by doBlockLevels()
1921          * @private
1922          */
1923         /* private */ function closeParagraph() {
1924                 $result = '';
1925                 if ( '' != $this->mLastSection ) {
1926                         $result = '</' . $this->mLastSection  . ">\n";
1927                 }
1928                 $this->mInPre = false;
1929                 $this->mLastSection = '';
1930                 return $result;
1931         }
1932         # getCommon() returns the length of the longest common substring
1933         # of both arguments, starting at the beginning of both.
1934         #
1935         /* private */ function getCommon( $st1, $st2 ) {
1936                 $fl = strlen( $st1 );
1937                 $shorter = strlen( $st2 );
1938                 if ( $fl < $shorter ) { $shorter = $fl; }
1939
1940                 for ( $i = 0; $i < $shorter; ++$i ) {
1941                         if ( $st1{$i} != $st2{$i} ) { break; }
1942                 }
1943                 return $i;
1944         }
1945         # These next three functions open, continue, and close the list
1946         # element appropriate to the prefix character passed into them.
1947         #
1948         /* private */ function openList( $char ) {
1949                 $result = $this->closeParagraph();
1950
1951                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1952                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1953                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1954                 else if ( ';' == $char ) {
1955                         $result .= '<dl><dt>';
1956                         $this->mDTopen = true;
1957                 }
1958                 else { $result = '<!-- ERR 1 -->'; }
1959
1960                 return $result;
1961         }
1962
1963         /* private */ function nextItem( $char ) {
1964                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1965                 else if ( ':' == $char || ';' == $char ) {
1966                         $close = '</dd>';
1967                         if ( $this->mDTopen ) { $close = '</dt>'; }
1968                         if ( ';' == $char ) {
1969                                 $this->mDTopen = true;
1970                                 return $close . '<dt>';
1971                         } else {
1972                                 $this->mDTopen = false;
1973                                 return $close . '<dd>';
1974                         }
1975                 }
1976                 return '<!-- ERR 2 -->';
1977         }
1978
1979         /* private */ function closeList( $char ) {
1980                 if ( '*' == $char ) { $text = '</li></ul>'; }
1981                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1982                 else if ( ':' == $char ) {
1983                         if ( $this->mDTopen ) {
1984                                 $this->mDTopen = false;
1985                                 $text = '</dt></dl>';
1986                         } else {
1987                                 $text = '</dd></dl>';
1988                         }
1989                 }
1990                 else {  return '<!-- ERR 3 -->'; }
1991                 return $text."\n";
1992         }
1993         /**#@-*/
1994
1995         /**
1996          * Make lists from lines starting with ':', '*', '#', etc.
1997          *
1998          * @private
1999          * @return string the lists rendered as HTML
2000          */
2001         function doBlockLevels( $text, $linestart ) {
2002                 $fname = 'Parser::doBlockLevels';
2003                 wfProfileIn( $fname );
2004
2005                 # Parsing through the text line by line.  The main thing
2006                 # happening here is handling of block-level elements p, pre,
2007                 # and making lists from lines starting with * # : etc.
2008                 #
2009                 $textLines = explode( "\n", $text );
2010
2011                 $lastPrefix = $output = '';
2012                 $this->mDTopen = $inBlockElem = false;
2013                 $prefixLength = 0;
2014                 $paragraphStack = false;
2015
2016                 if ( !$linestart ) {
2017                         $output .= array_shift( $textLines );
2018                 }
2019                 foreach ( $textLines as $oLine ) {
2020                         $lastPrefixLength = strlen( $lastPrefix );
2021                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2022                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2023                         if ( !$this->mInPre ) {
2024                                 # Multiple prefixes may abut each other for nested lists.
2025                                 $prefixLength = strspn( $oLine, '*#:;' );
2026                                 $pref = substr( $oLine, 0, $prefixLength );
2027
2028                                 # eh?
2029                                 $pref2 = str_replace( ';', ':', $pref );
2030                                 $t = substr( $oLine, $prefixLength );
2031                                 $this->mInPre = !empty($preOpenMatch);
2032                         } else {
2033                                 # Don't interpret any other prefixes in preformatted text
2034                                 $prefixLength = 0;
2035                                 $pref = $pref2 = '';
2036                                 $t = $oLine;
2037                         }
2038
2039                         # List generation
2040                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2041                                 # Same as the last item, so no need to deal with nesting or opening stuff
2042                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2043                                 $paragraphStack = false;
2044
2045                                 if ( substr( $pref, -1 ) == ';') {
2046                                         # The one nasty exception: definition lists work like this:
2047                                         # ; title : definition text
2048                                         # So we check for : in the remainder text to split up the
2049                                         # title and definition, without b0rking links.
2050                                         $term = $t2 = '';
2051                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2052                                                 $t = $t2;
2053                                                 $output .= $term . $this->nextItem( ':' );
2054                                         }
2055                                 }
2056                         } elseif( $prefixLength || $lastPrefixLength ) {
2057                                 # Either open or close a level...
2058                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2059                                 $paragraphStack = false;
2060
2061                                 while( $commonPrefixLength < $lastPrefixLength ) {
2062                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2063                                         --$lastPrefixLength;
2064                                 }
2065                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2066                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2067                                 }
2068                                 while ( $prefixLength > $commonPrefixLength ) {
2069                                         $char = substr( $pref, $commonPrefixLength, 1 );
2070                                         $output .= $this->openList( $char );
2071
2072                                         if ( ';' == $char ) {
2073                                                 # FIXME: This is dupe of code above
2074                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2075                                                         $t = $t2;
2076                                                         $output .= $term . $this->nextItem( ':' );
2077                                                 }
2078                                         }
2079                                         ++$commonPrefixLength;
2080                                 }
2081                                 $lastPrefix = $pref2;
2082                         }
2083                         if( 0 == $prefixLength ) {
2084                                 wfProfileIn( "$fname-paragraph" );
2085                                 # No prefix (not in list)--go to paragraph mode
2086                                 // XXX: use a stack for nestable elements like span, table and div
2087                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2088                                 $closematch = preg_match(
2089                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2090                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2091                                 if ( $openmatch or $closematch ) {
2092                                         $paragraphStack = false;
2093                                         # TODO bug 5718: paragraph closed
2094                                         $output .= $this->closeParagraph();
2095                                         if ( $preOpenMatch and !$preCloseMatch ) {
2096                                                 $this->mInPre = true;
2097                                         }
2098                                         if ( $closematch ) {
2099                                                 $inBlockElem = false;
2100                                         } else {
2101                                                 $inBlockElem = true;
2102                                         }
2103                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2104                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2105                                                 // pre
2106                                                 if ($this->mLastSection != 'pre') {
2107                                                         $paragraphStack = false;
2108                                                         $output .= $this->closeParagraph().'<pre>';
2109                                                         $this->mLastSection = 'pre';
2110                                                 }
2111                                                 $t = substr( $t, 1 );
2112                                         } else {
2113                                                 // paragraph
2114                                                 if ( '' == trim($t) ) {
2115                                                         if ( $paragraphStack ) {
2116                                                                 $output .= $paragraphStack.'<br />';
2117                                                                 $paragraphStack = false;
2118                                                                 $this->mLastSection = 'p';
2119                                                         } else {
2120                                                                 if ($this->mLastSection != 'p' ) {
2121                                                                         $output .= $this->closeParagraph();
2122                                                                         $this->mLastSection = '';
2123                                                                         $paragraphStack = '<p>';
2124                                                                 } else {
2125                                                                         $paragraphStack = '</p><p>';
2126                                                                 }
2127                                                         }
2128                                                 } else {
2129                                                         if ( $paragraphStack ) {
2130                                                                 $output .= $paragraphStack;
2131                                                                 $paragraphStack = false;
2132                                                                 $this->mLastSection = 'p';
2133                                                         } else if ($this->mLastSection != 'p') {
2134                                                                 $output .= $this->closeParagraph().'<p>';
2135                                                                 $this->mLastSection = 'p';
2136                                                         }
2137                                                 }
2138                                         }
2139                                 }
2140                                 wfProfileOut( "$fname-paragraph" );
2141                         }
2142                         // somewhere above we forget to get out of pre block (bug 785)
2143                         if($preCloseMatch && $this->mInPre) {
2144                                 $this->mInPre = false;
2145                         }
2146                         if ($paragraphStack === false) {
2147                                 $output .= $t."\n";
2148                         }
2149                 }
2150                 while ( $prefixLength ) {
2151                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2152                         --$prefixLength;
2153                 }
2154                 if ( '' != $this->mLastSection ) {
2155                         $output .= '</' . $this->mLastSection . '>';
2156                         $this->mLastSection = '';
2157                 }
2158
2159                 wfProfileOut( $fname );
2160                 return $output;
2161         }
2162
2163         /**
2164          * Split up a string on ':', ignoring any occurences inside tags
2165          * to prevent illegal overlapping.
2166          * @param string $str the string to split
2167          * @param string &$before set to everything before the ':'
2168          * @param string &$after set to everything after the ':'
2169          * return string the position of the ':', or false if none found
2170          */
2171         function findColonNoLinks($str, &$before, &$after) {
2172                 $fname = 'Parser::findColonNoLinks';
2173                 wfProfileIn( $fname );
2174
2175                 $pos = strpos( $str, ':' );
2176                 if( $pos === false ) {
2177                         // Nothing to find!
2178                         wfProfileOut( $fname );
2179                         return false;
2180                 }
2181
2182                 $lt = strpos( $str, '<' );
2183                 if( $lt === false || $lt > $pos ) {
2184                         // Easy; no tag nesting to worry about
2185                         $before = substr( $str, 0, $pos );
2186                         $after = substr( $str, $pos+1 );
2187                         wfProfileOut( $fname );
2188                         return $pos;
2189                 }
2190
2191                 // Ugly state machine to walk through avoiding tags.
2192                 $state = self::COLON_STATE_TEXT;
2193                 $stack = 0;
2194                 $len = strlen( $str );
2195                 for( $i = 0; $i < $len; $i++ ) {
2196                         $c = $str{$i};
2197
2198                         switch( $state ) {
2199                         // (Using the number is a performance hack for common cases)
2200                         case 0: // self::COLON_STATE_TEXT:
2201                                 switch( $c ) {
2202                                 case "<":
2203                                         // Could be either a <start> tag or an </end> tag
2204                                         $state = self::COLON_STATE_TAGSTART;
2205                                         break;
2206                                 case ":":
2207                                         if( $stack == 0 ) {
2208                                                 // We found it!
2209                                                 $before = substr( $str, 0, $i );
2210                                                 $after = substr( $str, $i + 1 );
2211                                                 wfProfileOut( $fname );
2212                                                 return $i;
2213                                         }
2214                                         // Embedded in a tag; don't break it.
2215                                         break;
2216                                 default:
2217                                         // Skip ahead looking for something interesting
2218                                         $colon = strpos( $str, ':', $i );
2219                                         if( $colon === false ) {
2220                                                 // Nothing else interesting
2221                                                 wfProfileOut( $fname );
2222                                                 return false;
2223                                         }
2224                                         $lt = strpos( $str, '<', $i );
2225                                         if( $stack === 0 ) {
2226                                                 if( $lt === false || $colon < $lt ) {
2227                                                         // We found it!
2228                                                         $before = substr( $str, 0, $colon );
2229                                                         $after = substr( $str, $colon + 1 );
2230                                                         wfProfileOut( $fname );
2231                                                         return $i;
2232                                                 }
2233                                         }
2234                                         if( $lt === false ) {
2235                                                 // Nothing else interesting to find; abort!
2236                                                 // We're nested, but there's no close tags left. Abort!
2237                                                 break 2;
2238                                         }
2239                                         // Skip ahead to next tag start
2240                                         $i = $lt;
2241                                         $state = self::COLON_STATE_TAGSTART;
2242                                 }
2243                                 break;
2244                         case 1: // self::COLON_STATE_TAG:
2245                                 // In a <tag>
2246                                 switch( $c ) {
2247                                 case ">":
2248                                         $stack++;
2249                                         $state = self::COLON_STATE_TEXT;
2250                                         break;
2251                                 case "/":
2252                                         // Slash may be followed by >?
2253                                         $state = self::COLON_STATE_TAGSLASH;
2254                                         break;
2255                                 default:
2256                                         // ignore
2257                                 }
2258                                 break;
2259                         case 2: // self::COLON_STATE_TAGSTART:
2260                                 switch( $c ) {
2261                                 case "/":
2262                                         $state = self::COLON_STATE_CLOSETAG;
2263                                         break;
2264                                 case "!":
2265                                         $state = self::COLON_STATE_COMMENT;
2266                                         break;
2267                                 case ">":
2268                                         // Illegal early close? This shouldn't happen D:
2269                                         $state = self::COLON_STATE_TEXT;
2270                                         break;
2271                                 default:
2272                                         $state = self::COLON_STATE_TAG;
2273                                 }
2274                                 break;
2275                         case 3: // self::COLON_STATE_CLOSETAG:
2276                                 // In a </tag>
2277                                 if( $c == ">" ) {
2278                                         $stack--;
2279                                         if( $stack < 0 ) {
2280                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2281                                                 wfProfileOut( $fname );
2282                                                 return false;
2283                                         }
2284                                         $state = self::COLON_STATE_TEXT;
2285                                 }
2286                                 break;
2287                         case self::COLON_STATE_TAGSLASH:
2288                                 if( $c == ">" ) {
2289                                         // Yes, a self-closed tag <blah/>
2290                                         $state = self::COLON_STATE_TEXT;
2291                                 } else {
2292                                         // Probably we're jumping the gun, and this is an attribute
2293                                         $state = self::COLON_STATE_TAG;
2294                                 }
2295                                 break;
2296                         case 5: // self::COLON_STATE_COMMENT:
2297                                 if( $c == "-" ) {
2298                                         $state = self::COLON_STATE_COMMENTDASH;
2299                                 }
2300                                 break;
2301                         case self::COLON_STATE_COMMENTDASH:
2302                                 if( $c == "-" ) {
2303                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2304                                 } else {
2305                                         $state = self::COLON_STATE_COMMENT;
2306                                 }
2307                                 break;
2308                         case self::COLON_STATE_COMMENTDASHDASH:
2309                                 if( $c == ">" ) {
2310                                         $state = self::COLON_STATE_TEXT;
2311                                 } else {
2312                                         $state = self::COLON_STATE_COMMENT;
2313                                 }
2314                                 break;
2315                         default:
2316                                 throw new MWException( "State machine error in $fname" );
2317                         }
2318                 }
2319                 if( $stack > 0 ) {
2320                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2321                         return false;
2322                 }
2323                 wfProfileOut( $fname );
2324                 return false;
2325         }
2326
2327         /**
2328          * Return value of a magic variable (like PAGENAME)
2329          *
2330          * @private
2331          */
2332         function getVariableValue( $index ) {
2333                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2334
2335                 /**
2336                  * Some of these require message or data lookups and can be
2337                  * expensive to check many times.
2338                  */
2339                 static $varCache = array();
2340                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2341                         if ( isset( $varCache[$index] ) ) {
2342                                 return $varCache[$index];
2343                         }
2344                 }
2345
2346                 $ts = time();
2347                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2348
2349                 # Use the time zone
2350                 global $wgLocaltimezone;
2351                 if ( isset( $wgLocaltimezone ) ) {
2352                         $oldtz = getenv( 'TZ' );
2353                         putenv( 'TZ='.$wgLocaltimezone );
2354                 }
2355
2356                 wfSuppressWarnings(); // E_STRICT system time bitching
2357                 $localTimestamp = date( 'YmdHis', $ts );
2358                 $localMonth = date( 'm', $ts );
2359                 $localMonthName = date( 'n', $ts );
2360                 $localDay = date( 'j', $ts );
2361                 $localDay2 = date( 'd', $ts );
2362                 $localDayOfWeek = date( 'w', $ts );
2363                 $localWeek = date( 'W', $ts );
2364                 $localYear = date( 'Y', $ts );
2365                 $localHour = date( 'H', $ts );
2366                 if ( isset( $wgLocaltimezone ) ) {
2367                         putenv( 'TZ='.$oldtz );
2368                 }
2369                 wfRestoreWarnings();
2370
2371                 switch ( $index ) {
2372                         case 'currentmonth':
2373                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2374                         case 'currentmonthname':
2375                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2376                         case 'currentmonthnamegen':
2377                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2378                         case 'currentmonthabbrev':
2379                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2380                         case 'currentday':
2381                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2382                         case 'currentday2':
2383                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2384                         case 'localmonth':
2385                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2386                         case 'localmonthname':
2387                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2388                         case 'localmonthnamegen':
2389                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2390                         case 'localmonthabbrev':
2391                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2392                         case 'localday':
2393                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2394                         case 'localday2':
2395                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2396                         case 'pagename':
2397                                 return wfEscapeWikiText( $this->mTitle->getText() );
2398                         case 'pagenamee':
2399                                 return $this->mTitle->getPartialURL();
2400                         case 'fullpagename':
2401                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2402                         case 'fullpagenamee':
2403                                 return $this->mTitle->getPrefixedURL();
2404                         case 'subpagename':
2405                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2406                         case 'subpagenamee':
2407                                 return $this->mTitle->getSubpageUrlForm();
2408                         case 'basepagename':
2409                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2410                         case 'basepagenamee':
2411                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2412                         case 'talkpagename':
2413                                 if( $this->mTitle->canTalk() ) {
2414                                         $talkPage = $this->mTitle->getTalkPage();
2415                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2416                                 } else {
2417                                         return '';
2418                                 }
2419                         case 'talkpagenamee':
2420                                 if( $this->mTitle->canTalk() ) {
2421                                         $talkPage = $this->mTitle->getTalkPage();
2422                                         return $talkPage->getPrefixedUrl();
2423                                 } else {
2424                                         return '';
2425                                 }
2426                         case 'subjectpagename':
2427                                 $subjPage = $this->mTitle->getSubjectPage();
2428                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2429                         case 'subjectpagenamee':
2430                                 $subjPage = $this->mTitle->getSubjectPage();
2431                                 return $subjPage->getPrefixedUrl();
2432                         case 'revisionid':
2433                                 // Let the edit saving system know we should parse the page
2434                                 // *after* a revision ID has been assigned.
2435                                 $this->mOutput->setFlag( 'vary-revision' );
2436                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2437                                 return $this->mRevisionId;
2438                         case 'revisionday':
2439                                 // Let the edit saving system know we should parse the page
2440                                 // *after* a revision ID has been assigned. This is for null edits.
2441                                 $this->mOutput->setFlag( 'vary-revision' );
2442                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2443                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2444                         case 'revisionday2':
2445                                 // Let the edit saving system know we should parse the page
2446                                 // *after* a revision ID has been assigned. This is for null edits.
2447                                 $this->mOutput->setFlag( 'vary-revision' );
2448                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2449                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2450                         case 'revisionmonth':
2451                                 // Let the edit saving system know we should parse the page
2452                                 // *after* a revision ID has been assigned. This is for null edits.
2453                                 $this->mOutput->setFlag( 'vary-revision' );
2454                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2455                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2456                         case 'revisionyear':
2457                                 // Let the edit saving system know we should parse the page
2458                                 // *after* a revision ID has been assigned. This is for null edits.
2459                                 $this->mOutput->setFlag( 'vary-revision' );
2460                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2461                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2462                         case 'revisiontimestamp':
2463                                 // Let the edit saving system know we should parse the page
2464                                 // *after* a revision ID has been assigned. This is for null edits.
2465                                 $this->mOutput->setFlag( 'vary-revision' );
2466                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2467                                 return $this->getRevisionTimestamp();
2468                         case 'namespace':
2469                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2470                         case 'namespacee':
2471                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2472                         case 'talkspace':
2473                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2474                         case 'talkspacee':
2475                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2476                         case 'subjectspace':
2477                                 return $this->mTitle->getSubjectNsText();
2478                         case 'subjectspacee':
2479                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2480                         case 'currentdayname':
2481                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2482                         case 'currentyear':
2483                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2484                         case 'currenttime':
2485                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2486                         case 'currenthour':
2487                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2488                         case 'currentweek':
2489                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2490                                 // int to remove the padding
2491                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2492                         case 'currentdow':
2493                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2494                         case 'localdayname':
2495                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2496                         case 'localyear':
2497                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2498                         case 'localtime':
2499                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2500                         case 'localhour':
2501                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2502                         case 'localweek':
2503                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2504                                 // int to remove the padding
2505                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2506                         case 'localdow':
2507                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2508                         case 'numberofarticles':
2509                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2510                         case 'numberoffiles':
2511                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2512                         case 'numberofusers':
2513                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2514                         case 'numberofpages':
2515                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2516                         case 'numberofadmins':
2517                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2518                         case 'numberofedits':
2519                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2520                         case 'currenttimestamp':
2521                                 return $varCache[$index] = wfTimestampNow();
2522                         case 'localtimestamp':
2523                                 return $varCache[$index] = $localTimestamp;
2524                         case 'currentversion':
2525                                 return $varCache[$index] = SpecialVersion::getVersion();
2526                         case 'sitename':
2527                                 return $wgSitename;
2528                         case 'server':
2529                                 return $wgServer;
2530                         case 'servername':
2531                                 return $wgServerName;
2532                         case 'scriptpath':
2533                                 return $wgScriptPath;
2534                         case 'directionmark':
2535                                 return $wgContLang->getDirMark();
2536                         case 'contentlanguage':
2537                                 global $wgContLanguageCode;
2538                                 return $wgContLanguageCode;
2539                         default:
2540                                 $ret = null;
2541                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2542                                         return $ret;
2543                                 else
2544                                         return null;
2545                 }
2546         }
2547
2548         /**
2549          * initialise the magic variables (like CURRENTMONTHNAME)
2550          *
2551          * @private
2552          */
2553         function initialiseVariables() {
2554                 $fname = 'Parser::initialiseVariables';
2555                 wfProfileIn( $fname );
2556                 $variableIDs = MagicWord::getVariableIDs();
2557
2558                 $this->mVariables = new MagicWordArray( $variableIDs );
2559                 wfProfileOut( $fname );
2560         }
2561
2562         /**
2563          * Preprocess some wikitext and return the document tree.
2564          * This is the ghost of replace_variables().
2565          *
2566          * @param string $text The text to parse
2567          * @param integer flags Bitwise combination of:
2568          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2569          *                                     included. Default is to assume a direct page view.
2570          *
2571          * The generated DOM tree must depend only on the input text, the flags, and $this->ot['msg'].
2572          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2573          *
2574          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2575          * change in the DOM tree for a given text, must be passed through the section identifier
2576          * in the section edit link and thus back to extractSections().
2577          *
2578          * The output of this function is currently only cached in process memory, but a persistent
2579          * cache may be implemented at a later date which takes further advantage of these strict
2580          * dependency requirements.
2581          *
2582          * @private
2583          */
2584         function preprocessToDom ( $text, $flags = 0 ) {
2585                 wfProfileIn( __METHOD__ );
2586                 wfProfileIn( __METHOD__.'-makexml' );
2587
2588                 static $msgRules, $normalRules, $inclusionSupertags, $nonInclusionSupertags;
2589                 if ( !$msgRules ) {
2590                         $msgRules = array(
2591                                 '{' => array(
2592                                         'end' => '}',
2593                                         'names' => array(
2594                                                 2 => 'template',
2595                                         ),
2596                                         'min' => 2,
2597                                         'max' => 2,
2598                                 ),
2599                                 '[' => array(
2600                                         'end' => ']',
2601                                         'names' => array( 2 => null ),
2602                                         'min' => 2,
2603                                         'max' => 2,
2604                                 )
2605                         );
2606                         $normalRules = array(
2607                                 '{' => array(
2608                                         'end' => '}',
2609                                         'names' => array(
2610                                                 2 => 'template',
2611                                                 3 => 'tplarg',
2612                                         ),
2613                                         'min' => 2,
2614                                         'max' => 3,
2615                                 ),
2616                                 '[' => array(
2617                                         'end' => ']',
2618                                         'names' => array( 2 => null ),
2619                                         'min' => 2,
2620                                         'max' => 2,
2621                                 )
2622                         );
2623                 }
2624                 if ( $this->ot['msg'] ) {
2625                         $rules = $msgRules;
2626                 } else {
2627                         $rules = $normalRules;
2628                 }
2629                 $forInclusion = $flags & self::PTD_FOR_INCLUSION;
2630
2631                 $xmlishElements = $this->getStripList();
2632                 $enableOnlyinclude = false;
2633                 if ( $forInclusion ) {
2634                         $ignoredTags = array( 'includeonly', '/includeonly' );
2635                         $ignoredElements = array( 'noinclude' );
2636                         $xmlishElements[] = 'noinclude';
2637                         if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
2638                                 $enableOnlyinclude = true;
2639                         }
2640                 } else {
2641                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
2642                         $ignoredElements = array( 'includeonly' );
2643                         $xmlishElements[] = 'includeonly';
2644                 }
2645                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
2646
2647                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2648                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
2649
2650                 $stack = array();      # Stack of unclosed parentheses
2651                 $stackIndex = -1;      # Stack read pointer
2652
2653                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2654                 $revText = strrev( $text ); // For fast reverse searches
2655
2656                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2657                 $topAccum = '<root>';      # Top level text accumulator
2658                 $accum =& $topAccum;            # Current text accumulator
2659                 $findEquals = false;            # True to find equals signs in arguments
2660                 $findHeading = false;           # True to look at LF characters for possible headings
2661                 $findPipe = false;              # True to take notice of pipe characters
2662                 $headingIndex = 1;
2663                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2664                 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
2665
2666                 if ( $enableOnlyinclude ) {
2667                         $i = 0;
2668                 }
2669
2670                 while ( true ) {
2671                         if ( $findOnlyinclude ) {
2672                                 // Ignore all input up to the next <onlyinclude>
2673                                 $startPos = strpos( $text, '<onlyinclude>', $i );
2674                                 if ( $startPos === false ) {
2675                                         // Ignored section runs to the end
2676                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
2677                                         break;
2678                                 }
2679                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
2680                                 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
2681                                 $i = $tagEndPos;
2682                                 $findOnlyinclude = false;
2683                         }
2684
2685                         if ( $i == -1 ) {
2686                                 $found = 'line-start';
2687                                 $curChar = '';
2688                         } else {
2689                                 # Find next opening brace, closing brace or pipe
2690                                 $search = $searchBase;
2691                                 if ( $stackIndex == -1 ) {
2692                                         $currentClosing = '';
2693                                         // Look for headings only at the top stack level
2694                                         // Among other things, this resolves the ambiguity between =
2695                                         // for headings and = for template arguments
2696                                         $search .= "\n";
2697                                 } else {
2698                                         $currentClosing = $stack[$stackIndex]['close'];
2699                                         $search .= $currentClosing;
2700                                 }
2701                                 if ( $findPipe ) {
2702                                         $search .= '|';
2703                                 }
2704                                 if ( $findEquals ) {
2705                                         $search .= '=';
2706                                 }
2707                                 $rule = null;
2708                                 # Output literal section, advance input counter
2709                                 $literalLength = strcspn( $text, $search, $i );
2710                                 if ( $literalLength > 0 ) {
2711                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2712                                         $i += $literalLength;
2713                                 }
2714                                 if ( $i >= strlen( $text ) ) {
2715                                         if ( $currentClosing == "\n" ) {
2716                                                 // Do a past-the-end run to finish off the heading
2717                                                 $curChar = '';
2718                                                 $found = 'line-end';
2719                                         } else {
2720                                                 # All done
2721                                                 break;
2722                                         }
2723                                 } else {
2724                                         $curChar = $text[$i];
2725                                         if ( $curChar == '|' ) {
2726                                                 $found = 'pipe';
2727                                         } elseif ( $curChar == '=' ) {
2728                                                 $found = 'equals';
2729                                         } elseif ( $curChar == '<' ) {
2730                                                 $found = 'angle';
2731                                         } elseif ( $curChar == "\n" ) {
2732                                                 if ( $stackIndex == -1 ) {
2733                                                         $found = 'line-start';
2734                                                 } else {
2735                                                         $found = 'line-end';
2736                                                 }
2737                                         } elseif ( $curChar == $currentClosing ) {
2738                                                 $found = 'close';
2739                                         } elseif ( isset( $rules[$curChar] ) ) {
2740                                                 $found = 'open';
2741                                                 $rule = $rules[$curChar];
2742                                         } else {
2743                                                 # Some versions of PHP have a strcspn which stops on null characters
2744                                                 # Ignore and continue
2745                                                 ++$i;
2746                                                 continue;
2747                                         }
2748                                 }
2749                         }
2750
2751                         if ( $found == 'angle' ) {
2752                                 $matches = false;
2753                                 // Handle </onlyinclude>
2754                                 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
2755                                         $findOnlyinclude = true;
2756                                         continue;
2757                                 }
2758
2759                                 // Determine element name
2760                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
2761                                         // Element name missing or not listed
2762                                         $accum .= '&lt;';
2763                                         ++$i;
2764                                         continue;
2765                                 }
2766                                 // Handle comments
2767                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2768                                         // To avoid leaving blank lines, when a comment is both preceded
2769                                         // and followed by a newline (ignoring spaces), trim leading and
2770                                         // trailing spaces and one of the newlines.
2771
2772                                         // Find the end
2773                                         $endPos = strpos( $text, '-->', $i + 4 );
2774                                         if ( $endPos === false ) {
2775                                                 // Unclosed comment in input, runs to end
2776                                                 $inner = substr( $text, $i );
2777                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2778                                                 $i = strlen( $text );
2779                                         } else {
2780                                                 // Search backwards for leading whitespace
2781                                                 $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
2782                                                 // Search forwards for trailing whitespace
2783                                                 // $wsEnd will be the position of the last space
2784                                                 $wsEnd = $endPos + 2 + strspn( $text, ' ', $endPos + 3 );
2785                                                 // Eat the line if possible
2786                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
2787                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n" )
2788                                                 {
2789                                                         $startPos = $wsStart;
2790                                                         $endPos = $wsEnd + 1;
2791                                                         // Remove leading whitespace from the end of the accumulator
2792                                                         // Sanity check first though
2793                                                         $wsLength = $i - $wsStart;
2794                                                         if ( $wsLength > 0 && substr( $accum, -$wsLength ) === str_repeat( ' ', $wsLength ) ) {
2795                                                                 $accum = substr( $accum, 0, -$wsLength );
2796                                                         }
2797                                                 } else {
2798                                                         // No line to eat, just take the comment itself
2799                                                         $startPos = $i;
2800                                                         $endPos += 2;
2801                                                 }
2802
2803                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
2804                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2805                                                 $i = $endPos + 1;
2806                                         }
2807                                         continue;
2808                                 }
2809                                 $name = $matches[1];
2810                                 $attrStart = $i + strlen( $name ) + 1;
2811
2812                                 // Find end of tag
2813                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2814                                 if ( $tagEndPos === false ) {
2815                                         // Infinite backtrack
2816                                         // Disable tag search to prevent worst-case O(N^2) performance
2817                                         $noMoreGT = true;
2818                                         $accum .= '&lt;';
2819                                         ++$i;
2820                                         continue;
2821                                 }
2822
2823                                 // Handle ignored tags
2824                                 if ( in_array( $name, $ignoredTags ) ) {
2825                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '</ignore>';
2826                                         $i = $tagEndPos + 1;
2827                                         continue;
2828                                 }
2829
2830                                 $tagStartPos = $i;
2831                                 if ( $text[$tagEndPos-1] == '/' ) {
2832                                         $attrEnd = $tagEndPos - 1;
2833                                         $inner = null;
2834                                         $i = $tagEndPos + 1;
2835                                         $close = '';
2836                                 } else {
2837                                         $attrEnd = $tagEndPos;
2838                                         // Find closing tag
2839                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2840                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2841                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2842                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2843                                         } else {
2844                                                 // No end tag -- let it run out to the end of the text.
2845                                                 $inner = substr( $text, $tagEndPos + 1 );
2846                                                 $i = strlen( $text );
2847                                                 $close = '';
2848                                         }
2849                                 }
2850                                 // <includeonly> and <noinclude> just become <ignore> tags
2851                                 if ( in_array( $name, $ignoredElements ) ) {
2852                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
2853                                                 . '</ignore>';
2854                                         continue;
2855                                 }
2856
2857                                 $accum .= '<ext>';
2858                                 if ( $attrEnd <= $attrStart ) {
2859                                         $attr = '';
2860                                 } else {
2861                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2862                                 }
2863                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2864                                         // Note that the attr element contains the whitespace between name and attribute,
2865                                         // this is necessary for precise reconstruction during pre-save transform.
2866                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2867                                 if ( $inner !== null ) {
2868                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2869                                 }
2870                                 $accum .= $close . '</ext>';
2871                         }
2872
2873                         elseif ( $found == 'line-start' ) {
2874                                 // Is this the start of a heading?
2875                                 // Line break belongs before the heading element in any case
2876                                 $accum .= $curChar;
2877                                 $i++;
2878
2879                                 $count = strspn( $text, '=', $i, 6 );
2880                                 if ( $count > 0 ) {
2881                                         $piece = array(
2882                                                 'open' => "\n",
2883                                                 'close' => "\n",
2884                                                 'parts' => array( str_repeat( '=', $count ) ),
2885                                                 'count' => $count );
2886                                         $stack[++$stackIndex] = $piece;
2887                                         $i += $count;
2888                                         $accum =& $stack[$stackIndex]['parts'][0];
2889                                         $findPipe = false;
2890                                 }
2891                         }
2892
2893                         elseif ( $found == 'line-end' ) {
2894                                 $piece = $stack[$stackIndex];
2895                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2896                                 assert( $piece['open'] == "\n" );
2897                                 assert( $stackIndex == 0 );
2898                                 // Search back through the input to see if it has a proper close
2899                                 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
2900                                 $m = false;
2901                                 $count = $piece['count'];
2902                                 if ( preg_match( "/\s*(={{$count}})/A", $revText, $m, 0, strlen( $text ) - $i ) ) {
2903                                         // Found match, output <h>
2904                                         $count = min( strlen( $m[1] ), $count );
2905                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2906                                         $headingIndex++;
2907                                 } else {
2908                                         // No match, no <h>, just pass down the inner text
2909                                         $element = $accum;
2910                                 }
2911                                 // Unwind the stack
2912                                 // Headings can only occur on the top level, so this is a bit simpler than the
2913                                 // generic stack unwind operation in the close case
2914                                 unset( $stack[$stackIndex--] );
2915                                 $accum =& $topAccum;
2916                                 $findEquals = false;
2917                                 $findPipe = false;
2918
2919                                 // Append the result to the enclosing accumulator
2920                                 $accum .= $element;
2921                                 // Note that we do NOT increment the input pointer.
2922                                 // This is because the closing linebreak could be the opening linebreak of
2923                                 // another heading. Infinite loops are avoided because the next iteration MUST
2924                                 // hit the heading open case above, which unconditionally increments the
2925                                 // input pointer.
2926                         }
2927
2928                         elseif ( $found == 'open' ) {
2929                                 # count opening brace characters
2930                                 $count = strspn( $text, $curChar, $i );
2931
2932                                 # we need to add to stack only if opening brace count is enough for one of the rules
2933                                 if ( $count >= $rule['min'] ) {
2934                                         # Add it to the stack
2935                                         $piece = array(
2936                                                 'open' => $curChar,
2937                                                 'close' => $rule['end'],
2938                                                 'count' => $count,
2939                                                 'parts' => array( '' ),
2940                                                 'eqpos' => array(),
2941                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2942                                         );
2943
2944                                         $stackIndex ++;
2945                                         $stack[$stackIndex] = $piece;
2946                                         $accum =& $stack[$stackIndex]['parts'][0];
2947                                         $findEquals = false;
2948                                         $findPipe = true;
2949                                 } else {
2950                                         # Add literal brace(s)
2951                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2952                                 }
2953                                 $i += $count;
2954                         }
2955
2956                         elseif ( $found == 'close' ) {
2957                                 $piece = $stack[$stackIndex];
2958                                 # lets check if there are enough characters for closing brace
2959                                 $maxCount = $piece['count'];
2960                                 $count = strspn( $text, $curChar, $i, $maxCount );
2961
2962                                 # check for maximum matching characters (if there are 5 closing
2963                                 # characters, we will probably need only 3 - depending on the rules)
2964                                 $matchingCount = 0;
2965                                 $rule = $rules[$piece['open']];
2966                                 if ( $count > $rule['max'] ) {
2967                                         # The specified maximum exists in the callback array, unless the caller
2968                                         # has made an error
2969                                         $matchingCount = $rule['max'];
2970                                 } else {
2971                                         # Count is less than the maximum
2972                                         # Skip any gaps in the callback array to find the true largest match
2973                                         # Need to use array_key_exists not isset because the callback can be null
2974                                         $matchingCount = $count;
2975                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2976                                                 --$matchingCount;
2977                                         }
2978                                 }
2979
2980                                 if ($matchingCount <= 0) {
2981                                         # No matching element found in callback array
2982                                         # Output a literal closing brace and continue
2983                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2984                                         $i += $count;
2985                                         continue;
2986                                 }
2987                                 $name = $rule['names'][$matchingCount];
2988                                 if ( $name === null ) {
2989                                         // No element, just literal text
2990                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2991                                                 implode( '|', $piece['parts'] ) .
2992                                                 str_repeat( $rule['end'], $matchingCount );
2993                                 } else {
2994                                         # Create XML element
2995                                         # Note: $parts is already XML, does not need to be encoded further
2996                                         $parts = $piece['parts'];
2997                                         $title = $parts[0];
2998                                         unset( $parts[0] );
2999
3000                                         # The invocation is at the start of the line if lineStart is set in
3001                                         # the stack, and all opening brackets are used up.
3002                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
3003                                                 $attr = ' lineStart="1"';
3004                                         } else {
3005                                                 $attr = '';
3006                                         }
3007
3008                                         $element = "<$name$attr>";
3009                                         $element .= "<title>$title</title>";
3010                                         $argIndex = 1;
3011                                         foreach ( $parts as $partIndex => $part ) {
3012                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
3013                                                         $eqpos = $piece['eqpos'][$partIndex];
3014                                                         list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
3015                                                         list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
3016                                                         $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
3017                                                 } else {
3018                                                         list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
3019                                                         $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
3020                                                         $argIndex++;
3021                                                 }
3022                                         }
3023                                         $element .= "</$name>";
3024                                 }
3025
3026                                 # Advance input pointer
3027                                 $i += $matchingCount;
3028
3029                                 # Unwind the stack
3030                                 unset( $stack[$stackIndex--] );
3031                                 if ( $stackIndex == -1 ) {
3032                                         $accum =& $topAccum;
3033                                         $findEquals = false;
3034                                         $findPipe = false;
3035                                 } else {
3036                                         $partCount = count( $stack[$stackIndex]['parts'] );
3037                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
3038                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
3039                                         $findEquals = $findPipe && $partCount > 1
3040                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
3041                                 }
3042
3043                                 # Re-add the old stack element if it still has unmatched opening characters remaining
3044                                 if ($matchingCount < $piece['count']) {
3045                                         $piece['parts'] = array( '' );
3046                                         $piece['count'] -= $matchingCount;
3047                                         $piece['eqpos'] = array();
3048                                         # do we still qualify for any callback with remaining count?
3049                                         $names = $rules[$piece['open']]['names'];
3050                                         $skippedBraces = 0;
3051                                         $enclosingAccum =& $accum;
3052                                         while ( $piece['count'] ) {
3053                                                 if ( array_key_exists( $piece['count'], $names ) ) {
3054                                                         $stackIndex++;
3055                                                         $stack[$stackIndex] = $piece;
3056                                                         $accum =& $stack[$stackIndex]['parts'][0];
3057                                                         $findEquals = true;
3058                                                         $findPipe = true;
3059                                                         break;
3060                                                 }
3061                                                 --$piece['count'];
3062                                                 $skippedBraces ++;
3063                                         }
3064                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
3065                                 }
3066
3067                                 # Add XML element to the enclosing accumulator
3068                                 $accum .= $element;
3069                         }
3070
3071                         elseif ( $found == 'pipe' ) {
3072                                 $stack[$stackIndex]['parts'][] = '';
3073                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3074                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
3075                                 $findEquals = true;
3076                                 ++$i;
3077                         }
3078
3079                         elseif ( $found == 'equals' ) {
3080                                 $findEquals = false;
3081                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3082                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
3083                                 $accum .= '=';
3084                                 ++$i;
3085                         }
3086                 }
3087
3088                 # Output any remaining unclosed brackets
3089                 foreach ( $stack as $piece ) {
3090                         if ( $piece['open'] == "\n" ) {
3091                                 $topAccum .= $piece['parts'][0];
3092                         } else {
3093                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
3094                         }
3095                 }
3096                 $topAccum .= '</root>';
3097
3098                 wfProfileOut( __METHOD__.'-makexml' );
3099                 wfProfileIn( __METHOD__.'-loadXML' );
3100                 $dom = new DOMDocument;
3101                 wfSuppressWarnings();
3102                 $result = $dom->loadXML( $topAccum );
3103                 wfRestoreWarnings();
3104                 if ( !$result ) {
3105                         // Try running the XML through UtfNormal to get rid of invalid characters
3106                         $topAccum = UtfNormal::cleanUp( $topAccum );
3107                         $result = $dom->loadXML( $topAccum );
3108                         if ( !$result ) {
3109                                 throw new MWException( __METHOD__.' generated invalid XML' );
3110                         }
3111                 }
3112                 wfProfileOut( __METHOD__.'-loadXML' );
3113                 wfProfileOut( __METHOD__ );
3114                 return $dom;
3115         }
3116
3117         /*
3118          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3119          */
3120         public static function splitWhitespace( $s ) {
3121                 $ltrimmed = ltrim( $s );
3122                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3123                 $trimmed = rtrim( $ltrimmed );
3124                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3125                 if ( $diff > 0 ) {
3126                         $w2 = substr( $ltrimmed, -$diff );
3127                 } else {
3128                         $w2 = '';
3129                 }
3130                 return array( $w1, $trimmed, $w2 );
3131         }
3132
3133         /**
3134          * Replace magic variables, templates, and template arguments
3135          * with the appropriate text. Templates are substituted recursively,
3136          * taking care to avoid infinite loops.
3137          *
3138          * Note that the substitution depends on value of $mOutputType:
3139          *  OT_WIKI: only {{subst:}} templates
3140          *  OT_MSG: only magic variables
3141          *  OT_HTML: all templates and magic variables
3142          *
3143          * @param string $tex The text to transform
3144          * @param PPFrame $frame Object describing the arguments passed to the template
3145          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3146          * @private
3147          */
3148         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3149                 # Prevent too big inclusions
3150                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3151                         return $text;
3152                 }
3153
3154                 $fname = __METHOD__;
3155                 wfProfileIn( $fname );
3156
3157                 if ( $frame === false ) {
3158                         $frame = new PPFrame( $this );
3159                 } elseif ( !( $frame instanceof PPFrame ) ) {
3160                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3161                 }
3162
3163                 $dom = $this->preprocessToDom( $text );
3164                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3165                 $text = $frame->expand( $dom, $flags );
3166
3167                 wfProfileOut( $fname );
3168                 return $text;
3169         }
3170
3171         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3172         static function createAssocArgs( $args ) {
3173                 $assocArgs = array();
3174                 $index = 1;
3175                 foreach( $args as $arg ) {
3176                         $eqpos = strpos( $arg, '=' );
3177                         if ( $eqpos === false ) {
3178                                 $assocArgs[$index++] = $arg;
3179                         } else {
3180                                 $name = trim( substr( $arg, 0, $eqpos ) );
3181                                 $value = trim( substr( $arg, $eqpos+1 ) );
3182                                 if ( $value === false ) {
3183                                         $value = '';
3184                                 }
3185                                 if ( $name !== false ) {
3186                                         $assocArgs[$name] = $value;
3187                                 }
3188                         }
3189                 }
3190
3191                 return $assocArgs;
3192         }
3193
3194         /**
3195          * Return the text of a template, after recursively
3196          * replacing any variables or templates within the template.
3197          *
3198          * @param array $piece The parts of the template
3199          *  $piece['text']: matched text
3200          *  $piece['title']: the title, i.e. the part before the |
3201          *  $piece['parts']: the parameter array
3202          * @param PPFrame The current frame, contains template arguments
3203          * @return string the text of the template
3204          * @private
3205          */
3206         function braceSubstitution( $piece, $frame ) {
3207                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3208                 $fname = __METHOD__;
3209                 wfProfileIn( $fname );
3210                 wfProfileIn( __METHOD__.'-setup' );
3211
3212                 # Flags
3213                 $found = false;             # $text has been filled
3214                 $nowiki = false;            # wiki markup in $text should be escaped
3215                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3216                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3217                 $isDOM = false;             # $text is a DOM node needing expansion
3218
3219                 # Title object, where $text came from
3220                 $title = NULL;
3221
3222                 # $part1 is the bit before the first |, and must contain only title characters.
3223                 # Various prefixes will be stripped from it later.
3224                 $titleWithSpaces = $frame->expand( $piece['title'] );
3225                 $part1 = trim( $titleWithSpaces );
3226                 $titleText = false;
3227
3228                 # Original title text preserved for various purposes
3229                 $originalTitle = $part1;
3230
3231                 # $args is a list of argument nodes, starting from index 0, not including $part1
3232                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3233                 wfProfileOut( __METHOD__.'-setup' );
3234
3235                 # SUBST
3236                 wfProfileIn( __METHOD__.'-modifiers' );
3237                 if ( !$found ) {
3238                         $mwSubst =& MagicWord::get( 'subst' );
3239                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3240                                 # One of two possibilities is true:
3241                                 # 1) Found SUBST but not in the PST phase
3242                                 # 2) Didn't find SUBST and in the PST phase
3243                                 # In either case, return without further processing
3244                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3245                                 $found = true;
3246                         }
3247                 }
3248
3249                 # Variables
3250                 if ( !$found && $args->length == 0 ) {
3251                         $id = $this->mVariables->matchStartToEnd( $part1 );
3252                         if ( $id !== false ) {
3253                                 $text = $this->getVariableValue( $id );
3254                                 $this->mOutput->mContainsOldMagic = true;
3255                                 $found = true;
3256                         }
3257                 }
3258
3259                 # MSG, MSGNW and RAW
3260                 if ( !$found ) {
3261                         # Check for MSGNW:
3262                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3263                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3264                                 $nowiki = true;
3265                         } else {
3266                                 # Remove obsolete MSG:
3267                                 $mwMsg =& MagicWord::get( 'msg' );
3268                                 $mwMsg->matchStartAndRemove( $part1 );
3269                         }
3270
3271                         # Check for RAW:
3272                         $mwRaw =& MagicWord::get( 'raw' );
3273                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3274                                 $forceRawInterwiki = true;
3275                         }
3276                 }
3277                 wfProfileOut( __METHOD__.'-modifiers' );
3278
3279                 # Save path level before recursing into functions & templates.
3280                 $lastPathLevel = $this->mTemplatePath;
3281
3282                 # Parser functions
3283                 if ( !$found ) {
3284                         wfProfileIn( __METHOD__ . '-pfunc' );
3285
3286                         $colonPos = strpos( $part1, ':' );
3287                         if ( $colonPos !== false ) {
3288                                 # Case sensitive functions
3289                                 $function = substr( $part1, 0, $colonPos );
3290                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3291                                         $function = $this->mFunctionSynonyms[1][$function];
3292                                 } else {
3293                                         # Case insensitive functions
3294                                         $function = strtolower( $function );
3295                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3296                                                 $function = $this->mFunctionSynonyms[0][$function];
3297                                         } else {
3298                                                 $function = false;
3299                                         }
3300                                 }
3301                                 if ( $function ) {
3302                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3303                                         $initialArgs = array( &$this );
3304                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3305                                         if ( $flags & SFH_OBJECT_ARGS ) {
3306                                                 # Add a frame parameter, and pass the arguments as an array
3307                                                 $allArgs = $initialArgs;
3308                                                 $allArgs[] = $frame;
3309                                                 foreach ( $args as $arg ) {
3310                                                         $funcArgs[] = $arg;
3311                                                 }
3312                                                 $allArgs[] = $funcArgs;
3313                                         } else {
3314                                                 # Convert arguments to plain text
3315                                                 foreach ( $args as $arg ) {
3316                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3317                                                 }
3318                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3319                                         }
3320
3321                                         $result = call_user_func_array( $callback, $allArgs );
3322                                         $found = true;
3323
3324                                         if ( is_array( $result ) ) {
3325                                                 if ( isset( $result[0] ) ) {
3326                                                         $text = $result[0];
3327                                                         unset( $result[0] );
3328                                                 }
3329
3330                                                 // Extract flags into the local scope
3331                                                 // This allows callers to set flags such as nowiki, found, etc.
3332                                                 extract( $result );
3333                                         } else {
3334                                                 $text = $result;
3335                                         }
3336                                 }
3337                         }
3338                         wfProfileOut( __METHOD__ . '-pfunc' );
3339                 }
3340
3341                 # Finish mangling title and then check for loops.
3342                 # Set $title to a Title object and $titleText to the PDBK
3343                 if ( !$found ) {
3344                         $ns = NS_TEMPLATE;
3345                         # Split the title into page and subpage
3346                         $subpage = '';
3347                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3348                         if ($subpage !== '') {
3349                                 $ns = $this->mTitle->getNamespace();
3350                         }
3351                         $title = Title::newFromText( $part1, $ns );
3352                         if ( $title ) {
3353                                 $titleText = $title->getPrefixedText();
3354                                 # Check for language variants if the template is not found
3355                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3356                                         $wgContLang->findVariantLink($part1, $title);
3357                                 }
3358                                 # Do infinite loop check
3359                                 if ( isset( $this->mTemplatePath[$titleText] ) ) {
3360                                         $found = true;
3361                                         $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
3362                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3363                                 }
3364                         }
3365                 }
3366
3367                 # Load from database
3368                 if ( !$found && $title ) {
3369                         wfProfileIn( __METHOD__ . '-loadtpl' );
3370                         if ( !$title->isExternal() ) {
3371                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3372                                         $text = SpecialPage::capturePath( $title );
3373                                         if ( is_string( $text ) ) {
3374                                                 $found = true;
3375                                                 $isHTML = true;
3376                                                 $this->disableCache();
3377                                         }
3378                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3379                                         $found = false; //access denied
3380                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3381                                 } else {
3382                                         list( $text, $title ) = $this->getTemplateDom( $title );
3383                                         if ( $text !== false ) {
3384                                                 $found = true;
3385                                                 $isDOM = true;
3386                                         }
3387                                 }
3388
3389                                 # If the title is valid but undisplayable, make a link to it
3390                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3391                                         $text = "[[:$titleText]]";
3392                                         $found = true;
3393                                 }
3394                         } elseif ( $title->isTrans() ) {
3395                                 // Interwiki transclusion
3396                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3397                                         $text = $this->interwikiTransclude( $title, 'render' );
3398                                         $isHTML = true;
3399                                 } else {
3400                                         $text = $this->interwikiTransclude( $title, 'raw' );
3401                                         // Preprocess it like a template
3402                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3403                                         $isDOM = true;
3404                                 }
3405                                 $found = true;
3406                         }
3407                         wfProfileOut( __METHOD__ . '-loadtpl' );
3408                 }
3409
3410                 # If we haven't found text to substitute by now, we're done
3411                 # Recover the source wikitext and return it
3412                 if ( !$found ) {
3413                         $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3414                         # Prune lower levels off the recursion check path
3415                         $this->mTemplatePath = $lastPathLevel;
3416                         wfProfileOut( $fname );
3417                         return $text;
3418                 }
3419
3420                 # Expand DOM-style return values in a child frame
3421                 if ( $isDOM ) {
3422                         # Clean up argument array
3423                         $newFrame = $frame->newChild( $args, $title );
3424                         # Add a new element to the templace recursion path
3425                         $this->mTemplatePath[$titleText] = 1;
3426
3427                         if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
3428                                 # Expansion is eligible for the empty-frame cache
3429                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3430                                         $text = $this->mTplExpandCache[$titleText];
3431                                 } else {
3432                                         $text = $newFrame->expand( $text );
3433                                         $this->mTplExpandCache[$titleText] = $text;
3434                                 }
3435                         } else {
3436                                 $text = $newFrame->expand( $text );
3437                         }
3438                 }
3439
3440                 # Replace raw HTML by a placeholder
3441                 # Add a blank line preceding, to prevent it from mucking up
3442                 # immediately preceding headings
3443                 if ( $isHTML ) {
3444                         $text = "\n\n" . $this->insertStripItem( $text );
3445                 }
3446                 # Escape nowiki-style return values
3447                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3448                         $text = wfEscapeWikiText( $text );
3449                 }
3450                 # Bug 529: if the template begins with a table or block-level
3451                 # element, it should be treated as beginning a new line.
3452                 # This behaviour is somewhat controversial.
3453                 elseif ( !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3454                         $text = "\n" . $text;
3455                 }
3456
3457                 # Prune lower levels off the recursion check path
3458                 $this->mTemplatePath = $lastPathLevel;
3459
3460                 if ( !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3461                         # Error, oversize inclusion
3462                         $text = "[[$originalTitle]]" .
3463                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3464                 }
3465
3466                 wfProfileOut( $fname );
3467                 return $text;
3468         }
3469
3470         /**
3471          * Get the semi-parsed DOM representation of a template with a given title,
3472          * and its redirect destination title. Cached.
3473          */
3474         function getTemplateDom( $title ) {
3475                 $cacheTitle = $title;
3476                 $titleText = $title->getPrefixedDBkey();
3477
3478                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3479                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3480                         $title = Title::makeTitle( $ns, $dbk );
3481                         $titleText = $title->getPrefixedDBkey();
3482                 }
3483                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3484                         return array( $this->mTplDomCache[$titleText], $title );
3485                 }
3486
3487                 // Cache miss, go to the database
3488                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3489
3490                 if ( $text === false ) {
3491                         $this->mTplDomCache[$titleText] = false;
3492                         return array( false, $title );
3493                 }
3494
3495                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3496                 $this->mTplDomCache[ $titleText ] = $dom;
3497
3498                 if (! $title->equals($cacheTitle)) {
3499                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3500                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3501                 }
3502
3503                 return array( $dom, $title );
3504         }
3505
3506         /**
3507          * Fetch the unparsed text of a template and register a reference to it.
3508          */
3509         function fetchTemplateAndTitle( $title ) {
3510                 $templateCb = $this->mOptions->getTemplateCallback();
3511                 $stuff = call_user_func( $templateCb, $title );
3512                 $text = $stuff['text'];
3513                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3514                 if ( isset( $stuff['deps'] ) ) {
3515                         foreach ( $stuff['deps'] as $dep ) {
3516                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3517                         }
3518                 }
3519                 return array($text,$finalTitle);
3520         }
3521
3522         function fetchTemplate( $title ) {
3523                 $rv = $this->fetchTemplateAndTitle($title);
3524                 return $rv[0];
3525         }
3526
3527         /**
3528          * Static function to get a template
3529          * Can be overridden via ParserOptions::setTemplateCallback().
3530          */
3531         static function statelessFetchTemplate( $title ) {
3532                 $text = $skip = false;
3533                 $finalTitle = $title;
3534                 $deps = array();
3535
3536                 // Loop to fetch the article, with up to 1 redirect
3537                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3538                         # Give extensions a chance to select the revision instead
3539                         $id = false; // Assume current
3540                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3541
3542                         if( $skip ) {
3543                                 $text = false;
3544                                 $deps[] = array(
3545                                         'title' => $title,
3546                                         'page_id' => $title->getArticleID(),
3547                                         'rev_id' => null );
3548                                 break;
3549                         }
3550                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3551                         $rev_id = $rev ? $rev->getId() : 0;
3552
3553                         $deps[] = array(
3554                                 'title' => $title,
3555                                 'page_id' => $title->getArticleID(),
3556                                 'rev_id' => $rev_id );
3557
3558                         if( $rev ) {
3559                                 $text = $rev->getText();
3560                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3561                                 global $wgLang;
3562                                 $message = $wgLang->lcfirst( $title->getText() );
3563                                 $text = wfMsgForContentNoTrans( $message );
3564                                 if( wfEmptyMsg( $message, $text ) ) {
3565                                         $text = false;
3566                                         break;
3567                                 }
3568                         } else {
3569                                 break;
3570                         }
3571                         if ( $text === false ) {
3572                                 break;
3573                         }
3574                         // Redirect?
3575                         $finalTitle = $title;
3576                         $title = Title::newFromRedirect( $text );
3577                 }
3578                 return array(
3579                         'text' => $text,
3580                         'finalTitle' => $finalTitle,
3581                         'deps' => $deps );
3582         }
3583
3584         /**
3585          * Transclude an interwiki link.
3586          */
3587         function interwikiTransclude( $title, $action ) {
3588                 global $wgEnableScaryTranscluding;
3589
3590                 if (!$wgEnableScaryTranscluding)
3591                         return wfMsg('scarytranscludedisabled');
3592
3593                 $url = $title->getFullUrl( "action=$action" );
3594
3595                 if (strlen($url) > 255)
3596                         return wfMsg('scarytranscludetoolong');
3597                 return $this->fetchScaryTemplateMaybeFromCache($url);
3598         }
3599
3600         function fetchScaryTemplateMaybeFromCache($url) {
3601                 global $wgTranscludeCacheExpiry;
3602                 $dbr = wfGetDB(DB_SLAVE);
3603                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3604                                 array('tc_url' => $url));
3605                 if ($obj) {
3606                         $time = $obj->tc_time;
3607                         $text = $obj->tc_contents;
3608                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3609                                 return $text;
3610                         }
3611                 }
3612
3613                 $text = Http::get($url);
3614                 if (!$text)
3615                         return wfMsg('scarytranscludefailed', $url);
3616
3617                 $dbw = wfGetDB(DB_MASTER);
3618                 $dbw->replace('transcache', array('tc_url'), array(
3619                         'tc_url' => $url,
3620                         'tc_time' => time(),
3621                         'tc_contents' => $text));
3622                 return $text;
3623         }
3624
3625
3626         /**
3627          * Triple brace replacement -- used for template arguments
3628          * @private
3629          */
3630         function argSubstitution( $piece, $frame ) {
3631                 wfProfileIn( __METHOD__ );
3632
3633                 $text = false;
3634                 $error = false;
3635                 $parts = $piece['parts'];
3636                 $argWithSpaces = $frame->expand( $piece['title'] );
3637                 $arg = trim( $argWithSpaces );
3638
3639                 if ( isset( $frame->args[$arg] ) ) {
3640                         $text = $frame->parent->expand( $frame->args[$arg] );
3641                 } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3642                         $text = $frame->expand( $parts->item( 0 ) );
3643                 }
3644                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3645                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3646                 }
3647
3648                 if ( $text === false ) {
3649                         $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
3650                 }
3651                 if ( $error !== false ) {
3652                         $text .= $error;
3653                 }
3654
3655                 wfProfileOut( __METHOD__ );
3656                 return $text;
3657         }
3658
3659         /**
3660          * Return the text to be used for a given extension tag.
3661          * This is the ghost of strip().
3662          *
3663          * @param array $params Associative array of parameters:
3664          *     name       DOMNode for the tag name
3665          *     attr       DOMNode for unparsed text where tag attributes are thought to be
3666          *     attributes Optional associative array of parsed attributes
3667          *     inner      Contents of extension element
3668          *     noClose    Original text did not have a close tag
3669          * @param PPFrame $frame
3670          */
3671         function extensionSubstitution( $params, $frame ) {
3672                 global $wgRawHtml, $wgContLang;
3673                 static $n = 1;
3674
3675                 $name = $frame->expand( $params['name'] );
3676                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3677                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3678
3679                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3680
3681                 if ( $this->ot['html'] ) {
3682                         $name = strtolower( $name );
3683
3684                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3685                         if ( isset( $params['attributes'] ) ) {
3686                                 $attributes = $attributes + $params['attributes'];
3687                         }
3688                         switch ( $name ) {
3689                                 case 'html':
3690                                         if( $wgRawHtml ) {
3691                                                 $output = $content;
3692                                                 break;
3693                                         } else {
3694                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3695                                         }
3696                                 case 'nowiki':
3697                                         $output = Xml::escapeTagsOnly( $content );
3698                                         break;
3699                                 case 'math':
3700                                         $output = $wgContLang->armourMath(
3701                                                 MathRenderer::renderMath( $content, $attributes ) );
3702                                         break;
3703                                 case 'gallery':
3704                                         $output = $this->renderImageGallery( $content, $attributes );
3705                                         break;
3706                                 default:
3707                                         if( isset( $this->mTagHooks[$name] ) ) {
3708                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3709                                                         array( $content, $attributes, $this ) );
3710                                         } else {
3711                                                 throw new MWException( "Invalid call hook $name" );
3712                                         }
3713                         }
3714                 } else {
3715                         if ( $content === null ) {
3716                                 $output = "<$name$attrText/>";
3717                         } else {
3718                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3719                                 $output = "<$name$attrText>$content$close";
3720                         }
3721                 }
3722
3723                 if ( $name == 'html' || $name == 'nowiki' ) {
3724                         $this->mStripState->nowiki->setPair( $marker, $output );
3725                 } else {
3726                         $this->mStripState->general->setPair( $marker, $output );
3727                 }
3728                 return $marker;
3729         }
3730
3731         /**
3732          * Increment an include size counter
3733          *
3734          * @param string $type The type of expansion
3735          * @param integer $size The size of the text
3736          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3737          */
3738         function incrementIncludeSize( $type, $size ) {
3739                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3740                         return false;
3741                 } else {
3742                         $this->mIncludeSizes[$type] += $size;
3743                         return true;
3744                 }
3745         }
3746
3747         /**
3748          * Detect __NOGALLERY__ magic word and set a placeholder
3749          */
3750         function stripNoGallery( &$text ) {
3751                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3752                 # do not add TOC
3753                 $mw = MagicWord::get( 'nogallery' );
3754                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3755         }
3756
3757         /**
3758          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3759          * placeholder that will then be replaced by the real TOC in
3760          * ->formatHeadings, this works because at this points real
3761          * comments will have already been discarded by the sanitizer.
3762          *
3763          * Any additional __TOC__ magic words left over will be discarded
3764          * as there can only be one TOC on the page.
3765          */
3766         function stripToc( $text ) {
3767                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3768                 # do not add TOC
3769                 $mw = MagicWord::get( 'notoc' );
3770                 if( $mw->matchAndRemove( $text ) ) {
3771                         $this->mShowToc = false;
3772                 }
3773
3774                 $mw = MagicWord::get( 'toc' );
3775                 if( $mw->match( $text ) ) {
3776                         $this->mShowToc = true;
3777                         $this->mForceTocPosition = true;
3778
3779                         // Set a placeholder. At the end we'll fill it in with the TOC.
3780                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3781
3782                         // Only keep the first one.
3783                         $text = $mw->replace( '', $text );
3784                 }
3785                 return $text;
3786         }
3787
3788         /**
3789          * This function accomplishes several tasks:
3790          * 1) Auto-number headings if that option is enabled
3791          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3792          * 3) Add a Table of contents on the top for users who have enabled the option
3793          * 4) Auto-anchor headings
3794          *
3795          * It loops through all headlines, collects the necessary data, then splits up the
3796          * string and re-inserts the newly formatted headlines.
3797          *
3798          * @param string $text
3799          * @param boolean $isMain
3800          * @private
3801          */
3802         function formatHeadings( $text, $isMain=true ) {
3803                 global $wgMaxTocLevel, $wgContLang;
3804
3805                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3806                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3807                         $showEditLink = 0;
3808                 } else {
3809                         $showEditLink = $this->mOptions->getEditSection();
3810                 }
3811
3812                 # Inhibit editsection links if requested in the page
3813                 $esw =& MagicWord::get( 'noeditsection' );
3814                 if( $esw->matchAndRemove( $text ) ) {
3815                         $showEditLink = 0;
3816                 }
3817
3818                 # Get all headlines for numbering them and adding funky stuff like [edit]
3819                 # links - this is for later, but we need the number of headlines right now
3820                 $matches = array();
3821                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3822
3823                 # if there are fewer than 4 headlines in the article, do not show TOC
3824                 # unless it's been explicitly enabled.
3825                 $enoughToc = $this->mShowToc &&
3826                         (($numMatches >= 4) || $this->mForceTocPosition);
3827
3828                 # Allow user to stipulate that a page should have a "new section"
3829                 # link added via __NEWSECTIONLINK__
3830                 $mw =& MagicWord::get( 'newsectionlink' );
3831                 if( $mw->matchAndRemove( $text ) )
3832                         $this->mOutput->setNewSection( true );
3833
3834                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3835                 # override above conditions and always show TOC above first header
3836                 $mw =& MagicWord::get( 'forcetoc' );
3837                 if ($mw->matchAndRemove( $text ) ) {
3838                         $this->mShowToc = true;
3839                         $enoughToc = true;
3840                 }
3841
3842                 # We need this to perform operations on the HTML
3843                 $sk = $this->mOptions->getSkin();
3844
3845                 # headline counter
3846                 $headlineCount = 0;
3847                 $numVisible = 0;
3848
3849                 # Ugh .. the TOC should have neat indentation levels which can be
3850                 # passed to the skin functions. These are determined here
3851                 $toc = '';
3852                 $full = '';
3853                 $head = array();
3854                 $sublevelCount = array();
3855                 $levelCount = array();
3856                 $toclevel = 0;
3857                 $level = 0;
3858                 $prevlevel = 0;
3859                 $toclevel = 0;
3860                 $prevtoclevel = 0;
3861                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3862                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3863                 $tocraw = array();
3864
3865                 foreach( $matches[3] as $headline ) {
3866                         $isTemplate = false;
3867                         $titleText = false;
3868                         $sectionIndex = false;
3869                         $numbering = '';
3870                         $markerMatches = array();
3871                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3872                                 $serial = $markerMatches[1];
3873                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3874                                 $isTemplate = ($titleText != $baseTitleText);
3875                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3876                         }
3877
3878                         if( $toclevel ) {
3879                                 $prevlevel = $level;
3880                                 $prevtoclevel = $toclevel;
3881                         }
3882                         $level = $matches[1][$headlineCount];
3883
3884                         if( $doNumberHeadings || $enoughToc ) {
3885
3886                                 if ( $level > $prevlevel ) {
3887                                         # Increase TOC level
3888                                         $toclevel++;
3889                                         $sublevelCount[$toclevel] = 0;
3890                                         if( $toclevel<$wgMaxTocLevel ) {
3891                                                 $prevtoclevel = $toclevel;
3892                                                 $toc .= $sk->tocIndent();
3893                                                 $numVisible++;
3894                                         }
3895                                 }
3896                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3897                                         # Decrease TOC level, find level to jump to
3898
3899                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3900                                                 # Can only go down to level 1
3901                                                 $toclevel = 1;
3902                                         } else {
3903                                                 for ($i = $toclevel; $i > 0; $i--) {
3904                                                         if ( $levelCount[$i] == $level ) {
3905                                                                 # Found last matching level
3906                                                                 $toclevel = $i;
3907                                                                 break;
3908                                                         }
3909                                                         elseif ( $levelCount[$i] < $level ) {
3910                                                                 # Found first matching level below current level
3911                                                                 $toclevel = $i + 1;
3912                                                                 break;
3913                                                         }
3914                                                 }
3915                                         }
3916                                         if( $toclevel<$wgMaxTocLevel ) {
3917                                                 if($prevtoclevel < $wgMaxTocLevel) {
3918                                                         # Unindent only if the previous toc level was shown :p
3919                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3920                                                 } else {
3921                                                         $toc .= $sk->tocLineEnd();
3922                                                 }
3923                                         }
3924                                 }
3925                                 else {
3926                                         # No change in level, end TOC line
3927                                         if( $toclevel<$wgMaxTocLevel ) {
3928                                                 $toc .= $sk->tocLineEnd();
3929                                         }
3930                                 }
3931
3932                                 $levelCount[$toclevel] = $level;
3933
3934                                 # count number of headlines for each level
3935                                 @$sublevelCount[$toclevel]++;
3936                                 $dot = 0;
3937                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3938                                         if( !empty( $sublevelCount[$i] ) ) {
3939                                                 if( $dot ) {
3940                                                         $numbering .= '.';
3941                                                 }
3942                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3943                                                 $dot = 1;
3944                                         }
3945                                 }
3946                         }
3947
3948                         # The safe header is a version of the header text safe to use for links
3949                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3950                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3951
3952                         # Remove link placeholders by the link text.
3953                         #     <!--LINK number-->
3954                         # turns into
3955                         #     link text with suffix
3956                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3957                                                             "\$this->mLinkHolders['texts'][\$1]",
3958                                                             $safeHeadline );
3959                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3960                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3961                                                             $safeHeadline );
3962
3963                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3964                         $tocline = preg_replace(
3965                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3966                                 array( '',                          '<$1>'),
3967                                 $safeHeadline
3968                         );
3969                         $tocline = trim( $tocline );
3970
3971                         # For the anchor, strip out HTML-y stuff period
3972                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3973                         $safeHeadline = trim( $safeHeadline );
3974
3975                         # Save headline for section edit hint before it's escaped
3976                         $headlineHint = $safeHeadline;
3977                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3978                         $refers[$headlineCount] = $safeHeadline;
3979
3980                         # count how many in assoc. array so we can track dupes in anchors
3981                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3982                         $refcount[$headlineCount] = $refers[$safeHeadline];
3983
3984                         # Don't number the heading if it is the only one (looks silly)
3985                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3986                                 # the two are different if the line contains a link
3987                                 $headline=$numbering . ' ' . $headline;
3988                         }
3989
3990                         # Create the anchor for linking from the TOC to the section
3991                         $anchor = $safeHeadline;
3992                         if($refcount[$headlineCount] > 1 ) {
3993                                 $anchor .= '_' . $refcount[$headlineCount];
3994                         }
3995                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3996                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3997                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3998                         }
3999                         # give headline the correct <h#> tag
4000                         if( $showEditLink && $sectionIndex !== false ) {
4001                                 if( $isTemplate ) {
4002                                         # Put a T flag in the section identifier, to indicate to extractSections()
4003                                         # that sections inside <includeonly> should be counted.
4004                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
4005                                 } else {
4006                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
4007                                 }
4008                         } else {
4009                                 $editlink = '';
4010                         }
4011                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
4012
4013                         $headlineCount++;
4014                 }
4015
4016                 $this->mOutput->setSections( $tocraw );
4017
4018                 # Never ever show TOC if no headers
4019                 if( $numVisible < 1 ) {
4020                         $enoughToc = false;
4021                 }
4022
4023                 if( $enoughToc ) {
4024                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4025                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
4026                         }
4027                         $toc = $sk->tocList( $toc );
4028                 }
4029
4030                 # split up and insert constructed headlines
4031
4032                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4033                 $i = 0;
4034
4035                 foreach( $blocks as $block ) {
4036                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
4037                                 # This is the [edit] link that appears for the top block of text when
4038                                 # section editing is enabled
4039
4040                                 # Disabled because it broke block formatting
4041                                 # For example, a bullet point in the top line
4042                                 # $full .= $sk->editSectionLink(0);
4043                         }
4044                         $full .= $block;
4045                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4046                                 # Top anchor now in skin
4047                                 $full = $full.$toc;
4048                         }
4049
4050                         if( !empty( $head[$i] ) ) {
4051                                 $full .= $head[$i];
4052                         }
4053                         $i++;
4054                 }
4055                 if( $this->mForceTocPosition ) {
4056                         return str_replace( '<!--MWTOC-->', $toc, $full );
4057                 } else {
4058                         return $full;
4059                 }
4060         }
4061
4062         /**
4063          * Transform wiki markup when saving a page by doing \r\n -> \n
4064          * conversion, substitting signatures, {{subst:}} templates, etc.
4065          *
4066          * @param string $text the text to transform
4067          * @param Title &$title the Title object for the current article
4068          * @param User &$user the User object describing the current user
4069          * @param ParserOptions $options parsing options
4070          * @param bool $clearState whether to clear the parser state first
4071          * @return string the altered wiki markup
4072          * @public
4073          */
4074         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
4075                 $this->mOptions = $options;
4076                 $this->mTitle =& $title;
4077                 $this->setOutputType( OT_WIKI );
4078
4079                 if ( $clearState ) {
4080                         $this->clearState();
4081                 }
4082
4083                 $pairs = array(
4084                         "\r\n" => "\n",
4085                 );
4086                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4087                 $text = $this->pstPass2( $text, $user );
4088                 $text = $this->mStripState->unstripBoth( $text );
4089                 return $text;
4090         }
4091
4092         /**
4093          * Pre-save transform helper function
4094          * @private
4095          */
4096         function pstPass2( $text, $user ) {
4097                 global $wgContLang, $wgLocaltimezone;
4098
4099                 /* Note: This is the timestamp saved as hardcoded wikitext to
4100                  * the database, we use $wgContLang here in order to give
4101                  * everyone the same signature and use the default one rather
4102                  * than the one selected in each user's preferences.
4103                  */
4104                 if ( isset( $wgLocaltimezone ) ) {
4105                         $oldtz = getenv( 'TZ' );
4106                         putenv( 'TZ='.$wgLocaltimezone );
4107                 }
4108                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4109                   ' (' . date( 'T' ) . ')';
4110                 if ( isset( $wgLocaltimezone ) ) {
4111                         putenv( 'TZ='.$oldtz );
4112                 }
4113
4114                 # Variable replacement
4115                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4116                 $text = $this->replaceVariables( $text );
4117
4118                 # Strip out <nowiki> etc. added via replaceVariables
4119                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4120
4121                 # Signatures
4122                 $sigText = $this->getUserSig( $user );
4123                 $text = strtr( $text, array(
4124                         '~~~~~' => $d,
4125                         '~~~~' => "$sigText $d",
4126                         '~~~' => $sigText
4127                 ) );
4128
4129                 # Context links: [[|name]] and [[name (context)|]]
4130                 #
4131                 global $wgLegalTitleChars;
4132                 $tc = "[$wgLegalTitleChars]";
4133                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4134
4135                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4136                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4137                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4138
4139                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4140                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4141                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4142
4143                 $t = $this->mTitle->getText();
4144                 $m = array();
4145                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4146                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4147                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4148                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4149                 } else {
4150                         # if there's no context, don't bother duplicating the title
4151                         $text = preg_replace( $p2, '[[\\1]]', $text );
4152                 }
4153
4154                 # Trim trailing whitespace
4155                 $text = rtrim( $text );
4156
4157                 return $text;
4158         }
4159
4160         /**
4161          * Fetch the user's signature text, if any, and normalize to
4162          * validated, ready-to-insert wikitext.
4163          *
4164          * @param User $user
4165          * @return string
4166          * @private
4167          */
4168         function getUserSig( &$user ) {
4169                 global $wgMaxSigChars;
4170
4171                 $username = $user->getName();
4172                 $nickname = $user->getOption( 'nickname' );
4173                 $nickname = $nickname === '' ? $username : $nickname;
4174
4175                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4176                         $nickname = $username;
4177                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4178                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4179                         # Sig. might contain markup; validate this
4180                         if( $this->validateSig( $nickname ) !== false ) {
4181                                 # Validated; clean up (if needed) and return it
4182                                 return $this->cleanSig( $nickname, true );
4183                         } else {
4184                                 # Failed to validate; fall back to the default
4185                                 $nickname = $username;
4186                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4187                         }
4188                 }
4189
4190                 // Make sure nickname doesnt get a sig in a sig
4191                 $nickname = $this->cleanSigInSig( $nickname );
4192
4193                 # If we're still here, make it a link to the user page
4194                 $userText = wfEscapeWikiText( $username );
4195                 $nickText = wfEscapeWikiText( $nickname );
4196                 if ( $user->isAnon() )  {
4197                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4198                 } else {
4199                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4200                 }
4201         }
4202
4203         /**
4204          * Check that the user's signature contains no bad XML
4205          *
4206          * @param string $text
4207          * @return mixed An expanded string, or false if invalid.
4208          */
4209         function validateSig( $text ) {
4210                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4211         }
4212
4213         /**
4214          * Clean up signature text
4215          *
4216          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4217          * 2) Substitute all transclusions
4218          *
4219          * @param string $text
4220          * @param $parsing Whether we're cleaning (preferences save) or parsing
4221          * @return string Signature text
4222          */
4223         function cleanSig( $text, $parsing = false ) {
4224                 if ( !$parsing ) {
4225                         global $wgTitle;
4226                         $this->startExternalParse( $wgTitle, new ParserOptions(), OT_MSG );
4227                 }
4228
4229                 # FIXME: regex doesn't respect extension tags or nowiki
4230                 #  => Move this logic to braceSubstitution()
4231                 $substWord = MagicWord::get( 'subst' );
4232                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4233                 $substText = '{{' . $substWord->getSynonym( 0 );
4234
4235                 $text = preg_replace( $substRegex, $substText, $text );
4236                 $text = $this->cleanSigInSig( $text );
4237                 $dom = $this->preprocessToDom( $text );
4238                 $frame = new PPFrame( $this );
4239                 $text = $frame->expand( $dom->documentElement );
4240
4241                 if ( !$parsing ) {
4242                         $text = $this->mStripState->unstripBoth( $text );
4243                 }
4244
4245                 return $text;
4246         }
4247
4248         /**
4249          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4250          * @param string $text
4251          * @return string Signature text with /~{3,5}/ removed
4252          */
4253         function cleanSigInSig( $text ) {
4254                 $text = preg_replace( '/~{3,5}/', '', $text );
4255                 return $text;
4256         }
4257
4258         /**
4259          * Set up some variables which are usually set up in parse()
4260          * so that an external function can call some class members with confidence
4261          * @public
4262          */
4263         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4264                 $this->mTitle =& $title;
4265                 $this->mOptions = $options;
4266                 $this->setOutputType( $outputType );
4267                 if ( $clearState ) {
4268                         $this->clearState();
4269                 }
4270         }
4271
4272         /**
4273          * Transform a MediaWiki message by replacing magic variables.
4274          *
4275          * For some unknown reason, it also expands templates, but only to the
4276          * first recursion level. This is wrong and broken, probably introduced
4277          * accidentally during refactoring, but probably relied upon by thousands
4278          * of users.
4279          *
4280          * @param string $text the text to transform
4281          * @param ParserOptions $options  options
4282          * @return string the text with variables substituted
4283          * @public
4284          */
4285         function transformMsg( $text, $options ) {
4286                 global $wgTitle;
4287                 static $executing = false;
4288
4289                 $fname = "Parser::transformMsg";
4290
4291                 # Guard against infinite recursion
4292                 if ( $executing ) {
4293                         return $text;
4294                 }
4295                 $executing = true;
4296
4297                 wfProfileIn($fname);
4298
4299                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4300                         $this->mTitle = $wgTitle;
4301                 } else {
4302                         $this->mTitle = Title::newFromText('msg');
4303                 }
4304                 $this->mOptions = $options;
4305                 $this->setOutputType( OT_MSG );
4306                 $this->clearState();
4307                 $text = $this->replaceVariables( $text );
4308                 $text = $this->mStripState->unstripBoth( $text );
4309
4310                 $executing = false;
4311                 wfProfileOut($fname);
4312                 return $text;
4313         }
4314
4315         /**
4316          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4317          * The callback should have the following form:
4318          *    function myParserHook( $text, $params, &$parser ) { ... }
4319          *
4320          * Transform and return $text. Use $parser for any required context, e.g. use
4321          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4322          *
4323          * @public
4324          *
4325          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4326          * @param mixed $callback The callback function (and object) to use for the tag
4327          *
4328          * @return The old value of the mTagHooks array associated with the hook
4329          */
4330         function setHook( $tag, $callback ) {
4331                 $tag = strtolower( $tag );
4332                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4333                 $this->mTagHooks[$tag] = $callback;
4334                 $this->mStripList[] = $tag;
4335
4336                 return $oldVal;
4337         }
4338
4339         function setTransparentTagHook( $tag, $callback ) {
4340                 $tag = strtolower( $tag );
4341                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4342                 $this->mTransparentTagHooks[$tag] = $callback;
4343
4344                 return $oldVal;
4345         }
4346
4347         /**
4348          * Create a function, e.g. {{sum:1|2|3}}
4349          * The callback function should have the form:
4350          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4351          *
4352          * The callback may either return the text result of the function, or an array with the text
4353          * in element 0, and a number of flags in the other elements. The names of the flags are
4354          * specified in the keys. Valid flags are:
4355          *   found                     The text returned is valid, stop processing the template. This
4356          *                             is on by default.
4357          *   nowiki                    Wiki markup in the return value should be escaped
4358          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4359          *   noargs                    Don't replace triple-brace arguments in the return value
4360          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4361          *
4362          * @public
4363          *
4364          * @param string $id The magic word ID
4365          * @param mixed $callback The callback function (and object) to use
4366          * @param integer $flags a combination of the following flags:
4367          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4368          *
4369          * @return The old callback function for this name, if any
4370          */
4371         function setFunctionHook( $id, $callback, $flags = 0 ) {
4372                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4373                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4374
4375                 # Add to function cache
4376                 $mw = MagicWord::get( $id );
4377                 if( !$mw )
4378                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4379
4380                 $synonyms = $mw->getSynonyms();
4381                 $sensitive = intval( $mw->isCaseSensitive() );
4382
4383                 foreach ( $synonyms as $syn ) {
4384                         # Case
4385                         if ( !$sensitive ) {
4386                                 $syn = strtolower( $syn );
4387                         }
4388                         # Add leading hash
4389                         if ( !( $flags & SFH_NO_HASH ) ) {
4390                                 $syn = '#' . $syn;
4391                         }
4392                         # Remove trailing colon
4393                         if ( substr( $syn, -1, 1 ) == ':' ) {
4394                                 $syn = substr( $syn, 0, -1 );
4395                         }
4396                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4397                 }
4398                 return $oldVal;
4399         }
4400
4401         /**
4402          * Get all registered function hook identifiers
4403          *
4404          * @return array
4405          */
4406         function getFunctionHooks() {
4407                 return array_keys( $this->mFunctionHooks );
4408         }
4409
4410         /**
4411          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4412          * Placeholders created in Skin::makeLinkObj()
4413          * Returns an array of link CSS classes, indexed by PDBK.
4414          * $options is a bit field, RLH_FOR_UPDATE to select for update
4415          */
4416         function replaceLinkHolders( &$text, $options = 0 ) {
4417                 global $wgUser;
4418                 global $wgContLang;
4419
4420                 $fname = 'Parser::replaceLinkHolders';
4421                 wfProfileIn( $fname );
4422
4423                 $pdbks = array();
4424                 $colours = array();
4425                 $linkcolour_ids = array();
4426                 $sk = $this->mOptions->getSkin();
4427                 $linkCache =& LinkCache::singleton();
4428
4429                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4430                         wfProfileIn( $fname.'-check' );
4431                         $dbr = wfGetDB( DB_SLAVE );
4432                         $page = $dbr->tableName( 'page' );
4433                         $threshold = $wgUser->getOption('stubthreshold');
4434
4435                         # Sort by namespace
4436                         asort( $this->mLinkHolders['namespaces'] );
4437
4438                         # Generate query
4439                         $query = false;
4440                         $current = null;
4441                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4442                                 # Make title object
4443                                 $title = $this->mLinkHolders['titles'][$key];
4444
4445                                 # Skip invalid entries.
4446                                 # Result will be ugly, but prevents crash.
4447                                 if ( is_null( $title ) ) {
4448                                         continue;
4449                                 }
4450                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4451
4452                                 # Check if it's a static known link, e.g. interwiki
4453                                 if ( $title->isAlwaysKnown() ) {
4454                                         $colours[$pdbk] = '';
4455                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4456                                         $colours[$pdbk] = '';
4457                                         $this->mOutput->addLink( $title, $id );
4458                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4459                                         $colours[$pdbk] = 'new';
4460                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4461                                         $colours[$pdbk] = 'new';
4462                                 } else {
4463                                         # Not in the link cache, add it to the query
4464                                         if ( !isset( $current ) ) {
4465                                                 $current = $ns;
4466                                                 $query =  "SELECT page_id, page_namespace, page_title";
4467                                                 if ( $threshold > 0 ) {
4468                                                         $query .= ', page_len, page_is_redirect';
4469                                                 }
4470                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4471                                         } elseif ( $current != $ns ) {
4472                                                 $current = $ns;
4473                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4474                                         } else {
4475                                                 $query .= ', ';
4476                                         }
4477
4478                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4479                                 }
4480                         }
4481                         if ( $query ) {
4482                                 $query .= '))';
4483                                 if ( $options & RLH_FOR_UPDATE ) {
4484                                         $query .= ' FOR UPDATE';
4485                                 }
4486
4487                                 $res = $dbr->query( $query, $fname );
4488
4489                                 # Fetch data and form into an associative array
4490                                 # non-existent = broken
4491                                 while ( $s = $dbr->fetchObject($res) ) {
4492                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4493                                         $pdbk = $title->getPrefixedDBkey();
4494                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4495                                         $this->mOutput->addLink( $title, $s->page_id );
4496                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4497                                         //add id to the extension todolist
4498                                         $linkcolour_ids[$s->page_id] = $pdbk;
4499                                 }
4500                                 //pass an array of page_ids to an extension
4501                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4502                         }
4503                         wfProfileOut( $fname.'-check' );
4504
4505                         # Do a second query for different language variants of links and categories
4506                         if($wgContLang->hasVariants()){
4507                                 $linkBatch = new LinkBatch();
4508                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4509                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4510                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4511
4512                                 $categories = $this->mOutput->getCategoryLinks();
4513
4514                                 // Add variants of links to link batch
4515                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4516                                         $title = $this->mLinkHolders['titles'][$key];
4517                                         if ( is_null( $title ) )
4518                                                 continue;
4519
4520                                         $pdbk = $title->getPrefixedDBkey();
4521                                         $titleText = $title->getText();
4522
4523                                         // generate all variants of the link title text
4524                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4525
4526                                         // if link was not found (in first query), add all variants to query
4527                                         if ( !isset($colours[$pdbk]) ){
4528                                                 foreach($allTextVariants as $textVariant){
4529                                                         if($textVariant != $titleText){
4530                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4531                                                                 if(is_null($variantTitle)) continue;
4532                                                                 $linkBatch->addObj( $variantTitle );
4533                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4534                                                         }
4535                                                 }
4536                                         }
4537                                 }
4538
4539                                 // process categories, check if a category exists in some variant
4540                                 foreach( $categories as $category ){
4541                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4542                                         foreach($variants as $variant){
4543                                                 if($variant != $category){
4544                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4545                                                         if(is_null($variantTitle)) continue;
4546                                                         $linkBatch->addObj( $variantTitle );
4547                                                         $categoryMap[$variant] = $category;
4548                                                 }
4549                                         }
4550                                 }
4551
4552
4553                                 if(!$linkBatch->isEmpty()){
4554                                         // construct query
4555                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4556
4557                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4558                                         if ( $threshold > 0 ) {
4559                                                 $variantQuery .= ', page_len, page_is_redirect';
4560                                         }
4561
4562                                         $variantQuery .= " FROM $page WHERE $titleClause";
4563                                         if ( $options & RLH_FOR_UPDATE ) {
4564                                                 $variantQuery .= ' FOR UPDATE';
4565                                         }
4566
4567                                         $varRes = $dbr->query( $variantQuery, $fname );
4568
4569                                         // for each found variants, figure out link holders and replace
4570                                         while ( $s = $dbr->fetchObject($varRes) ) {
4571
4572                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4573                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4574                                                 $vardbk = $variantTitle->getDBkey();
4575
4576                                                 $holderKeys = array();
4577                                                 if(isset($variantMap[$varPdbk])){
4578                                                         $holderKeys = $variantMap[$varPdbk];
4579                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4580                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4581                                                 }
4582
4583                                                 // loop over link holders
4584                                                 foreach($holderKeys as $key){
4585                                                         $title = $this->mLinkHolders['titles'][$key];
4586                                                         if ( is_null( $title ) ) continue;
4587
4588                                                         $pdbk = $title->getPrefixedDBkey();
4589
4590                                                         if(!isset($colours[$pdbk])){
4591                                                                 // found link in some of the variants, replace the link holder data
4592                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4593                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4594
4595                                                                 // set pdbk and colour
4596                                                                 $pdbks[$key] = $varPdbk;
4597                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4598                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4599                                                         }
4600                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4601                                                 }
4602
4603                                                 // check if the object is a variant of a category
4604                                                 if(isset($categoryMap[$vardbk])){
4605                                                         $oldkey = $categoryMap[$vardbk];
4606                                                         if($oldkey != $vardbk)
4607                                                                 $varCategories[$oldkey]=$vardbk;
4608                                                 }
4609                                         }
4610
4611                                         // rebuild the categories in original order (if there are replacements)
4612                                         if(count($varCategories)>0){
4613                                                 $newCats = array();
4614                                                 $originalCats = $this->mOutput->getCategories();
4615                                                 foreach($originalCats as $cat => $sortkey){
4616                                                         // make the replacement
4617                                                         if( array_key_exists($cat,$varCategories) )
4618                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4619                                                         else $newCats[$cat] = $sortkey;
4620                                                 }
4621                                                 $this->mOutput->setCategoryLinks($newCats);
4622                                         }
4623                                 }
4624                         }
4625
4626                         # Construct search and replace arrays
4627                         wfProfileIn( $fname.'-construct' );
4628                         $replacePairs = array();
4629                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4630                                 $pdbk = $pdbks[$key];
4631                                 $searchkey = "<!--LINK $key-->";
4632                                 $title = $this->mLinkHolders['titles'][$key];
4633                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4634                                         $linkCache->addBadLinkObj( $title );
4635                                         $colours[$pdbk] = 'new';
4636                                         $this->mOutput->addLink( $title, 0 );
4637                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4638                                                                         $this->mLinkHolders['texts'][$key],
4639                                                                         $this->mLinkHolders['queries'][$key] );
4640                                 } else {
4641                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4642                                                                         $this->mLinkHolders['texts'][$key],
4643                                                                         $this->mLinkHolders['queries'][$key] );
4644                                 }
4645                         }
4646                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4647                         wfProfileOut( $fname.'-construct' );
4648
4649                         # Do the thing
4650                         wfProfileIn( $fname.'-replace' );
4651                         $text = preg_replace_callback(
4652                                 '/(<!--LINK .*?-->)/',
4653                                 $replacer->cb(),
4654                                 $text);
4655
4656                         wfProfileOut( $fname.'-replace' );
4657                 }
4658
4659                 # Now process interwiki link holders
4660                 # This is quite a bit simpler than internal links
4661                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4662                         wfProfileIn( $fname.'-interwiki' );
4663                         # Make interwiki link HTML
4664                         $replacePairs = array();
4665                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4666                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4667                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4668                         }
4669                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4670
4671                         $text = preg_replace_callback(
4672                                 '/<!--IWLINK (.*?)-->/',
4673                                 $replacer->cb(),
4674                                 $text );
4675                         wfProfileOut( $fname.'-interwiki' );
4676                 }
4677
4678                 wfProfileOut( $fname );
4679                 return $colours;
4680         }
4681
4682         /**
4683          * Replace <!--LINK--> link placeholders with plain text of links
4684          * (not HTML-formatted).
4685          * @param string $text
4686          * @return string
4687          */
4688         function replaceLinkHoldersText( $text ) {
4689                 $fname = 'Parser::replaceLinkHoldersText';
4690                 wfProfileIn( $fname );
4691
4692                 $text = preg_replace_callback(
4693                         '/<!--(LINK|IWLINK) (.*?)-->/',
4694                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4695                         $text );
4696
4697                 wfProfileOut( $fname );
4698                 return $text;
4699         }
4700
4701         /**
4702          * @param array $matches
4703          * @return string
4704          * @private
4705          */
4706         function replaceLinkHoldersTextCallback( $matches ) {
4707                 $type = $matches[1];
4708                 $key  = $matches[2];
4709                 if( $type == 'LINK' ) {
4710                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4711                                 return $this->mLinkHolders['texts'][$key];
4712                         }
4713                 } elseif( $type == 'IWLINK' ) {
4714                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4715                                 return $this->mInterwikiLinkHolders['texts'][$key];
4716                         }
4717                 }
4718                 return $matches[0];
4719         }
4720
4721         /**
4722          * Tag hook handler for 'pre'.
4723          */
4724         function renderPreTag( $text, $attribs ) {
4725                 // Backwards-compatibility hack
4726                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4727
4728                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4729                 return wfOpenElement( 'pre', $attribs ) .
4730                         Xml::escapeTagsOnly( $content ) .
4731                         '</pre>';
4732         }
4733
4734         /**
4735          * Renders an image gallery from a text with one line per image.
4736          * text labels may be given by using |-style alternative text. E.g.
4737          *   Image:one.jpg|The number "1"
4738          *   Image:tree.jpg|A tree
4739          * given as text will return the HTML of a gallery with two images,
4740          * labeled 'The number "1"' and
4741          * 'A tree'.
4742          */
4743         function renderImageGallery( $text, $params ) {
4744                 $ig = new ImageGallery();
4745                 $ig->setContextTitle( $this->mTitle );
4746                 $ig->setShowBytes( false );
4747                 $ig->setShowFilename( false );
4748                 $ig->setParser( $this );
4749                 $ig->setHideBadImages();
4750                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4751                 $ig->useSkin( $this->mOptions->getSkin() );
4752                 $ig->mRevisionId = $this->mRevisionId;
4753
4754                 if( isset( $params['caption'] ) ) {
4755                         $caption = $params['caption'];
4756                         $caption = htmlspecialchars( $caption );
4757                         $caption = $this->replaceInternalLinks( $caption );
4758                         $ig->setCaptionHtml( $caption );
4759                 }
4760                 if( isset( $params['perrow'] ) ) {
4761                         $ig->setPerRow( $params['perrow'] );
4762                 }
4763                 if( isset( $params['widths'] ) ) {
4764                         $ig->setWidths( $params['widths'] );
4765                 }
4766                 if( isset( $params['heights'] ) ) {
4767                         $ig->setHeights( $params['heights'] );
4768                 }
4769
4770                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4771
4772                 $lines = explode( "\n", $text );
4773                 foreach ( $lines as $line ) {
4774                         # match lines like these:
4775                         # Image:someimage.jpg|This is some image
4776                         $matches = array();
4777                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4778                         # Skip empty lines
4779                         if ( count( $matches ) == 0 ) {
4780                                 continue;
4781                         }
4782                         $tp = Title::newFromText( $matches[1] );
4783                         $nt =& $tp;
4784                         if( is_null( $nt ) ) {
4785                                 # Bogus title. Ignore these so we don't bomb out later.
4786                                 continue;
4787                         }
4788                         if ( isset( $matches[3] ) ) {
4789                                 $label = $matches[3];
4790                         } else {
4791                                 $label = '';
4792                         }
4793
4794                         $pout = $this->parse( $label,
4795                                 $this->mTitle,
4796                                 $this->mOptions,
4797                                 false, // Strip whitespace...?
4798                                 false  // Don't clear state!
4799                         );
4800                         $html = $pout->getText();
4801
4802                         $ig->add( $nt, $html );
4803
4804                         # Only add real images (bug #5586)
4805                         if ( $nt->getNamespace() == NS_IMAGE ) {
4806                                 $this->mOutput->addImage( $nt->getDBkey() );
4807                         }
4808                 }
4809                 return $ig->toHTML();
4810         }
4811
4812         function getImageParams( $handler ) {
4813                 if ( $handler ) {
4814                         $handlerClass = get_class( $handler );
4815                 } else {
4816                         $handlerClass = '';
4817                 }
4818                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4819                         // Initialise static lists
4820                         static $internalParamNames = array(
4821                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4822                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4823                                         'bottom', 'text-bottom' ),
4824                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4825                                         'upright', 'border' ),
4826                         );
4827                         static $internalParamMap;
4828                         if ( !$internalParamMap ) {
4829                                 $internalParamMap = array();
4830                                 foreach ( $internalParamNames as $type => $names ) {
4831                                         foreach ( $names as $name ) {
4832                                                 $magicName = str_replace( '-', '_', "img_$name" );
4833                                                 $internalParamMap[$magicName] = array( $type, $name );
4834                                         }
4835                                 }
4836                         }
4837
4838                         // Add handler params
4839                         $paramMap = $internalParamMap;
4840                         if ( $handler ) {
4841                                 $handlerParamMap = $handler->getParamMap();
4842                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4843                                         $paramMap[$magic] = array( 'handler', $paramName );
4844                                 }
4845                         }
4846                         $this->mImageParams[$handlerClass] = $paramMap;
4847                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4848                 }
4849                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4850         }
4851
4852         /**
4853          * Parse image options text and use it to make an image
4854          */
4855         function makeImage( $title, $options ) {
4856                 # @TODO: let the MediaHandler specify its transform parameters
4857                 #
4858                 # Check if the options text is of the form "options|alt text"
4859                 # Options are:
4860                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4861                 #  * left               no resizing, just left align. label is used for alt= only
4862                 #  * right              same, but right aligned
4863                 #  * none               same, but not aligned
4864                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4865                 #  * center             center the image
4866                 #  * framed             Keep original image size, no magnify-button.
4867                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4868                 #  * upright            reduce width for upright images, rounded to full __0 px
4869                 #  * border             draw a 1px border around the image
4870                 # vertical-align values (no % or length right now):
4871                 #  * baseline
4872                 #  * sub
4873                 #  * super
4874                 #  * top
4875                 #  * text-top
4876                 #  * middle
4877                 #  * bottom
4878                 #  * text-bottom
4879
4880                 $parts = array_map( 'trim', explode( '|', $options) );
4881                 $sk = $this->mOptions->getSkin();
4882
4883                 # Give extensions a chance to select the file revision for us
4884                 $skip = $time = false;
4885                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4886
4887                 if ( $skip ) {
4888                         return $sk->makeLinkObj( $title );
4889                 }
4890
4891                 # Get parameter map
4892                 $file = wfFindFile( $title, $time );
4893                 $handler = $file ? $file->getHandler() : false;
4894
4895                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4896
4897                 # Process the input parameters
4898                 $caption = '';
4899                 $params = array( 'frame' => array(), 'handler' => array(),
4900                         'horizAlign' => array(), 'vertAlign' => array() );
4901                 foreach( $parts as $part ) {
4902                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4903                         if ( isset( $paramMap[$magicName] ) ) {
4904                                 list( $type, $paramName ) = $paramMap[$magicName];
4905                                 $params[$type][$paramName] = $value;
4906
4907                                 // Special case; width and height come in one variable together
4908                                 if( $type == 'handler' && $paramName == 'width' ) {
4909                                         $m = array();
4910                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4911                                                 $params[$type]['width'] = intval( $m[1] );
4912                                                 $params[$type]['height'] = intval( $m[2] );
4913                                         } else {
4914                                                 $params[$type]['width'] = intval( $value );
4915                                         }
4916                                 }
4917                         } else {
4918                                 $caption = $part;
4919                         }
4920                 }
4921
4922                 # Process alignment parameters
4923                 if ( $params['horizAlign'] ) {
4924                         $params['frame']['align'] = key( $params['horizAlign'] );
4925                 }
4926                 if ( $params['vertAlign'] ) {
4927                         $params['frame']['valign'] = key( $params['vertAlign'] );
4928                 }
4929
4930                 # Validate the handler parameters
4931                 if ( $handler ) {
4932                         foreach ( $params['handler'] as $name => $value ) {
4933                                 if ( !$handler->validateParam( $name, $value ) ) {
4934                                         unset( $params['handler'][$name] );
4935                                 }
4936                         }
4937                 }
4938
4939                 # Strip bad stuff out of the alt text
4940                 $alt = $this->replaceLinkHoldersText( $caption );
4941
4942                 # make sure there are no placeholders in thumbnail attributes
4943                 # that are later expanded to html- so expand them now and
4944                 # remove the tags
4945                 $alt = $this->mStripState->unstripBoth( $alt );
4946                 $alt = Sanitizer::stripAllTags( $alt );
4947
4948                 $params['frame']['alt'] = $alt;
4949                 $params['frame']['caption'] = $caption;
4950
4951                 # Linker does the rest
4952                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4953
4954                 # Give the handler a chance to modify the parser object
4955                 if ( $handler ) {
4956                         $handler->parserTransformHook( $this, $file );
4957                 }
4958
4959                 return $ret;
4960         }
4961
4962         /**
4963          * Set a flag in the output object indicating that the content is dynamic and
4964          * shouldn't be cached.
4965          */
4966         function disableCache() {
4967                 wfDebug( "Parser output marked as uncacheable.\n" );
4968                 $this->mOutput->mCacheTime = -1;
4969         }
4970
4971         /**#@+
4972          * Callback from the Sanitizer for expanding items found in HTML attribute
4973          * values, so they can be safely tested and escaped.
4974          * @param string $text
4975          * @param PPFrame $frame
4976          * @return string
4977          * @private
4978          */
4979         function attributeStripCallback( &$text, $frame = false ) {
4980                 $text = $this->replaceVariables( $text, $frame );
4981                 $text = $this->mStripState->unstripBoth( $text );
4982                 return $text;
4983         }
4984
4985         /**#@-*/
4986
4987         /**#@+
4988          * Accessor/mutator
4989          */
4990         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4991         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4992         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4993         /**#@-*/
4994
4995         /**#@+
4996          * Accessor
4997          */
4998         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4999         /**#@-*/
5000
5001
5002         /**
5003          * Break wikitext input into sections, and either pull or replace
5004          * some particular section's text.
5005          *
5006          * External callers should use the getSection and replaceSection methods.
5007          *
5008          * @param string $text Page wikitext
5009          * @param string $section A section identifier string of the form:
5010          *   <flag1> - <flag2> - ... - <section number>
5011          *
5012          * Currently the only recognised flag is "T", which means the target section number
5013          * was derived during a template inclusion parse, in other words this is a template
5014          * section edit link. If no flags are given, it was an ordinary section edit link.
5015          * This flag is required to avoid a section numbering mismatch when a section is
5016          * enclosed by <includeonly> (bug 6563).
5017          *
5018          * The section number 0 pulls the text before the first heading; other numbers will
5019          * pull the given section along with its lower-level subsections. If the section is
5020          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5021          *
5022          * @param string $mode One of "get" or "replace"
5023          * @param string $newText Replacement text for section data.
5024          * @return string for "get", the extracted section text.
5025          *                for "replace", the whole page with the section replaced.
5026          */
5027         private function extractSections( $text, $section, $mode, $newText='' ) {
5028                 global $wgTitle;
5029                 $this->clearState();
5030                 $this->mTitle = $wgTitle; // not generally used but removes an ugly failure mode
5031                 $this->mOptions = new ParserOptions;
5032                 $this->setOutputType( OT_WIKI );
5033                 $curIndex = 0;
5034                 $outText = '';
5035                 $frame = new PPFrame( $this );
5036
5037                 // Process section extraction flags
5038                 $flags = 0;
5039                 $sectionParts = explode( '-', $section );
5040                 $sectionIndex = array_pop( $sectionParts );
5041                 foreach ( $sectionParts as $part ) {
5042                         if ( $part == 'T' ) {
5043                                 $flags |= self::PTD_FOR_INCLUSION;
5044                         }
5045                 }
5046                 // Preprocess the text
5047                 $dom = $this->preprocessToDom( $text, $flags );
5048                 $root = $dom->documentElement;
5049
5050                 // <h> nodes indicate section breaks
5051                 // They can only occur at the top level, so we can find them by iterating the root's children
5052                 $node = $root->firstChild;
5053
5054                 // Find the target section
5055                 if ( $sectionIndex == 0 ) {
5056                         // Section zero doesn't nest, level=big
5057                         $targetLevel = 1000;
5058                 } else {
5059                         while ( $node ) {
5060                                 if ( $node->nodeName == 'h' ) {
5061                                         if ( $curIndex + 1 == $sectionIndex ) {
5062                                                 break;
5063                                         }
5064                                         $curIndex++;
5065                                 }
5066                                 if ( $mode == 'replace' ) {
5067                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5068                                 }
5069                                 $node = $node->nextSibling;
5070                         }
5071                         if ( $node ) {
5072                                 $targetLevel = $node->getAttribute( 'level' );
5073                         }
5074                 }
5075
5076                 if ( !$node ) {
5077                         // Not found
5078                         if ( $mode == 'get' ) {
5079                                 return $newText;
5080                         } else {
5081                                 return $text;
5082                         }
5083                 }
5084
5085                 // Find the end of the section, including nested sections
5086                 do {
5087                         if ( $node->nodeName == 'h' ) {
5088                                 $curIndex++;
5089                                 $curLevel = $node->getAttribute( 'level' );
5090                                 if ( $curIndex != $sectionIndex && $curLevel <= $targetLevel ) {
5091                                         break;
5092                                 }
5093                         }
5094                         if ( $mode == 'get' ) {
5095                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5096                         }
5097                         $node = $node->nextSibling;
5098                 } while ( $node );
5099
5100                 // Write out the remainder (in replace mode only)
5101                 if ( $mode == 'replace' ) {
5102                         // Output the replacement text
5103                         // Add two newlines on -- trailing whitespace in $newText is conventionally
5104                         // stripped by the editor, so we need both newlines to restore the paragraph gap
5105                         $outText .= $newText . "\n\n";
5106                         while ( $node ) {
5107                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5108                                 $node = $node->nextSibling;
5109                         }
5110                 }
5111
5112                 if ( is_string( $outText ) ) {
5113                         // Re-insert stripped tags
5114                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5115                 }
5116
5117                 return $outText;
5118         }
5119
5120         /**
5121          * This function returns the text of a section, specified by a number ($section).
5122          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5123          * the first section before any such heading (section 0).
5124          *
5125          * If a section contains subsections, these are also returned.
5126          *
5127          * @param string $text text to look in
5128          * @param string $section section identifier
5129          * @param string $deftext default to return if section is not found
5130          * @return string text of the requested section
5131          */
5132         public function getSection( $text, $section, $deftext='' ) {
5133                 return $this->extractSections( $text, $section, "get", $deftext );
5134         }
5135
5136         public function replaceSection( $oldtext, $section, $text ) {
5137                 return $this->extractSections( $oldtext, $section, "replace", $text );
5138         }
5139
5140         /**
5141          * Get the timestamp associated with the current revision, adjusted for
5142          * the default server-local timestamp
5143          */
5144         function getRevisionTimestamp() {
5145                 if ( is_null( $this->mRevisionTimestamp ) ) {
5146                         wfProfileIn( __METHOD__ );
5147                         global $wgContLang;
5148                         $dbr = wfGetDB( DB_SLAVE );
5149                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5150                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5151
5152                         // Normalize timestamp to internal MW format for timezone processing.
5153                         // This has the added side-effect of replacing a null value with
5154                         // the current time, which gives us more sensible behavior for
5155                         // previews.
5156                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5157
5158                         // The cryptic '' timezone parameter tells to use the site-default
5159                         // timezone offset instead of the user settings.
5160                         //
5161                         // Since this value will be saved into the parser cache, served
5162                         // to other users, and potentially even used inside links and such,
5163                         // it needs to be consistent for all visitors.
5164                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5165
5166                         wfProfileOut( __METHOD__ );
5167                 }
5168                 return $this->mRevisionTimestamp;
5169         }
5170
5171         /**
5172          * Mutator for $mDefaultSort
5173          *
5174          * @param $sort New value
5175          */
5176         public function setDefaultSort( $sort ) {
5177                 $this->mDefaultSort = $sort;
5178         }
5179
5180         /**
5181          * Accessor for $mDefaultSort
5182          * Will use the title/prefixed title if none is set
5183          *
5184          * @return string
5185          */
5186         public function getDefaultSort() {
5187                 if( $this->mDefaultSort !== false ) {
5188                         return $this->mDefaultSort;
5189                 } else {
5190                         return $this->mTitle->getNamespace() == NS_CATEGORY
5191                                         ? $this->mTitle->getText()
5192                                         : $this->mTitle->getPrefixedText();
5193                 }
5194         }
5195
5196         /**
5197          * Try to guess the section anchor name based on a wikitext fragment
5198          * presumably extracted from a heading, for example "Header" from
5199          * "== Header ==".
5200          */
5201         public function guessSectionNameFromWikiText( $text ) {
5202                 # Strip out wikitext links(they break the anchor)
5203                 $text = $this->stripSectionName( $text );
5204                 $headline = Sanitizer::decodeCharReferences( $text );
5205                 # strip out HTML
5206                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5207                 $headline = trim( $headline );
5208                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5209                 $replacearray = array(
5210                         '%3A' => ':',
5211                         '%' => '.'
5212                 );
5213                 return str_replace(
5214                         array_keys( $replacearray ),
5215                         array_values( $replacearray ),
5216                         $sectionanchor );
5217         }
5218
5219         /**
5220          * Strips a text string of wikitext for use in a section anchor
5221          *
5222          * Accepts a text string and then removes all wikitext from the
5223          * string and leaves only the resultant text (i.e. the result of
5224          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5225          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5226          * to create valid section anchors by mimicing the output of the
5227          * parser when headings are parsed.
5228          *
5229          * @param $text string Text string to be stripped of wikitext
5230          * for use in a Section anchor
5231          * @return Filtered text string
5232          */
5233         public function stripSectionName( $text ) {
5234                 # Strip internal link markup
5235                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5236                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5237
5238                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5239                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5240                 # on how many empty links there are on the page - need to figure that out.
5241                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5242
5243                 # Parse wikitext quotes (italics & bold)
5244                 $text = $this->doQuotes($text);
5245
5246                 # Strip HTML tags
5247                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5248                 return $text;
5249         }
5250
5251         /**
5252          * strip/replaceVariables/unstrip for preprocessor regression testing
5253          */
5254         function srvus( $text ) {
5255                 $text = $this->replaceVariables( $text );
5256                 $text = $this->mStripState->unstripBoth( $text );
5257                 $text = Sanitizer::removeHTMLtags( $text );
5258                 return $text;
5259         }
5260 }
5261
5262 /**
5263  * @todo document, briefly.
5264  * @addtogroup Parser
5265  */
5266 class OnlyIncludeReplacer {
5267         var $output = '';
5268
5269         function replace( $matches ) {
5270                 if ( substr( $matches[1], -1 ) == "\n" ) {
5271                         $this->output .= substr( $matches[1], 0, -1 );
5272                 } else {
5273                         $this->output .= $matches[1];
5274                 }
5275         }
5276 }
5277
5278 /**
5279  * @todo document, briefly.
5280  * @addtogroup Parser
5281  */
5282 class StripState {
5283         var $general, $nowiki;
5284
5285         function __construct() {
5286                 $this->general = new ReplacementArray;
5287                 $this->nowiki = new ReplacementArray;
5288         }
5289
5290         function unstripGeneral( $text ) {
5291                 wfProfileIn( __METHOD__ );
5292                 do {
5293                         $oldText = $text;
5294                         $text = $this->general->replace( $text );
5295                 } while ( $text != $oldText );
5296                 wfProfileOut( __METHOD__ );
5297                 return $text;
5298         }
5299
5300         function unstripNoWiki( $text ) {
5301                 wfProfileIn( __METHOD__ );
5302                 do {
5303                         $oldText = $text;
5304                         $text = $this->nowiki->replace( $text );
5305                 } while ( $text != $oldText );
5306                 wfProfileOut( __METHOD__ );
5307                 return $text;
5308         }
5309
5310         function unstripBoth( $text ) {
5311                 wfProfileIn( __METHOD__ );
5312                 do {
5313                         $oldText = $text;
5314                         $text = $this->general->replace( $text );
5315                         $text = $this->nowiki->replace( $text );
5316                 } while ( $text != $oldText );
5317                 wfProfileOut( __METHOD__ );
5318                 return $text;
5319         }
5320 }
5321
5322 /**
5323  * An expansion frame, used as a context to expand the result of preprocessToDom()
5324  */
5325 class PPFrame {
5326         var $parser, $title;
5327         var $titleCache;
5328
5329         const NO_ARGS = 1;
5330         const NO_TEMPLATES = 2;
5331         const STRIP_COMMENTS = 4;
5332         const NO_IGNORE = 8;
5333
5334         const RECOVER_ORIG = 11;
5335
5336         /**
5337          * Construct a new preprocessor frame.
5338          * @param Parser $parser The parent parser
5339          * @param Title $title The context title, or false if there isn't one
5340          */
5341         function __construct( $parser ) {
5342                 $this->parser = $parser;
5343                 $this->title = $parser->mTitle;
5344                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
5345         }
5346
5347         /**
5348          * Create a new child frame
5349          * $args is optionally a DOMNodeList containing the template arguments
5350          */
5351         function newChild( $args = false, $title = false ) {
5352                 $assocArgs = array();
5353                 if ( $title === false ) {
5354                         $title = $this->title;
5355                 }
5356                 if ( $args !== false ) {
5357                         $xpath = false;
5358                         foreach ( $args as $arg ) {
5359                                 if ( !$xpath ) {
5360                                         $xpath = new DOMXPath( $arg->ownerDocument );
5361                                 }
5362
5363                                 $nameNodes = $xpath->query( 'name', $arg );
5364                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5365                                         // Numbered parameter
5366                                         $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5367                                 } else {
5368                                         // Named parameter
5369                                         $name = $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS );
5370                                 }
5371
5372                                 $value = $xpath->query( 'value', $arg );
5373                                 $assocArgs[$name] = $value->item( 0 );
5374                         }
5375                 }
5376                 return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
5377         }
5378
5379         /**
5380          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5381          * using the current context
5382          * @param $root the node
5383          */
5384         function expand( $root, $flags = 0 ) {
5385                 if ( is_string( $root ) ) {
5386                         return $root;
5387                 }
5388
5389                 if ( $this->parser->ot['html']
5390                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5391                 {
5392                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5393                 }
5394
5395                 if ( is_array( $root ) ) {
5396                         $s = '';
5397                         foreach ( $root as $node ) {
5398                                 $s .= $this->expand( $node, $flags );
5399                         }
5400                 } elseif ( $root instanceof DOMNodeList ) {
5401                         $s = '';
5402                         foreach ( $root as $node ) {
5403                                 $s .= $this->expand( $node, $flags );
5404                         }
5405                 } elseif ( $root instanceof DOMNode ) {
5406                         if ( $root->nodeType == XML_TEXT_NODE ) {
5407                                 $s = $root->nodeValue;
5408                         } elseif ( $root->nodeName == 'template' ) {
5409                                 # Double-brace expansion
5410                                 $xpath = new DOMXPath( $root->ownerDocument );
5411                                 $titles = $xpath->query( 'title', $root );
5412                                 $title = $titles->item( 0 );
5413                                 $parts = $xpath->query( 'part', $root );
5414                                 if ( $flags & self::NO_TEMPLATES ) {
5415                                         $s = '{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}';
5416                                 } else {
5417                                         $lineStart = $root->getAttribute( 'lineStart' );
5418                                         $params = array(
5419                                                 'title' => $title,
5420                                                 'parts' => $parts,
5421                                                 'lineStart' => $lineStart,
5422                                                 'text' => 'FIXME' );
5423                                         $s = $this->parser->braceSubstitution( $params, $this );
5424                                 }
5425                         } elseif ( $root->nodeName == 'tplarg' ) {
5426                                 # Triple-brace expansion
5427                                 $xpath = new DOMXPath( $root->ownerDocument );
5428                                 $titles = $xpath->query( 'title', $root );
5429                                 $title = $titles->item( 0 );
5430                                 $parts = $xpath->query( 'part', $root );
5431                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5432                                         $s = '{{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}}';
5433                                 } else {
5434                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5435                                         $s = $this->parser->argSubstitution( $params, $this );
5436                                 }
5437                         } elseif ( $root->nodeName == 'comment' ) {
5438                                 # HTML-style comment
5439                                 if ( $this->parser->ot['html']
5440                                         || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
5441                                         || ( $flags & self::STRIP_COMMENTS ) )
5442                                 {
5443                                         $s = '';
5444                                 } else {
5445                                         $s = $root->textContent;
5446                                 }
5447                         } elseif ( $root->nodeName == 'ignore' ) {
5448                                 # Output suppression used by <includeonly> etc.
5449                                 # OT_WIKI will only respect <ignore> in substed templates.
5450                                 # The other output types respect it unless NO_IGNORE is set.
5451                                 # extractSections() sets NO_IGNORE and so never respects it.
5452                                 if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & self::NO_IGNORE ) ) {
5453                                         $s = $root->textContent;
5454                                 } else {
5455                                         $s = '';
5456                                 }
5457                         } elseif ( $root->nodeName == 'ext' ) {
5458                                 # Extension tag
5459                                 $xpath = new DOMXPath( $root->ownerDocument );
5460                                 $names = $xpath->query( 'name', $root );
5461                                 $attrs = $xpath->query( 'attr', $root );
5462                                 $inners = $xpath->query( 'inner', $root );
5463                                 $closes = $xpath->query( 'close', $root );
5464                                 $params = array(
5465                                         'name' => $names->item( 0 ),
5466                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5467                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5468                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5469                                 );
5470                                 $s = $this->parser->extensionSubstitution( $params, $this );
5471                         } elseif ( $root->nodeName == 'h' ) {
5472                                 # Heading
5473                                 $s = $this->expand( $root->childNodes, $flags );
5474
5475                                 if ( $this->parser->ot['html'] ) {
5476                                         # Insert heading index marker
5477                                         $headingIndex = $root->getAttribute( 'i' );
5478                                         $titleText = $this->title->getPrefixedDBkey();
5479                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5480                                         $serial = count( $this->parser->mHeadings ) - 1;
5481                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5482                                         $count = $root->getAttribute( 'level' );
5483                                         $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5484                                         $this->parser->mStripState->general->setPair( $marker, '' );
5485                                 }
5486                         } else {
5487                                 # Generic recursive expansion
5488                                 $s = '';
5489                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5490                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5491                                                 $s .= $node->nodeValue;
5492                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5493                                                 $s .= $this->expand( $node, $flags );
5494                                         }
5495                                 }
5496                         }
5497                 } else {
5498                         throw new MWException( __METHOD__.': Invalid parameter type' );
5499                 }
5500                 return $s;
5501         }
5502
5503         function implodeWithFlags( $sep, $flags /*, ... */ ) {
5504                 $args = array_slice( func_get_args(), 2 );
5505
5506                 $first = true;
5507                 $s = '';
5508                 foreach ( $args as $root ) {
5509                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5510                                 $root = array( $root );
5511                         }
5512                         foreach ( $root as $node ) {
5513                                 if ( $first ) {
5514                                         $first = false;
5515                                 } else {
5516                                         $s .= $sep;
5517                                 }
5518                                 $s .= $this->expand( $node, $flags );
5519                         }
5520                 }
5521                 return $s;
5522         }
5523
5524         function implode( $sep /*, ... */ ) {
5525                 $args = func_get_args();
5526                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0 ), array_slice( $args, 1 ) );
5527                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5528         }
5529
5530         /**
5531          * Split an <arg> or <template> node into a three-element array:
5532          *    DOMNode name, string index and DOMNode value
5533          */
5534         function splitBraceNode( $node ) {
5535                 $xpath = new DOMXPath( $node->ownerDocument );
5536                 $names = $xpath->query( 'name', $node );
5537                 $values = $xpath->query( 'value', $node );
5538                 if ( !$names->length || !$values->length ) {
5539                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
5540                 }
5541                 $name = $names->item( 0 );
5542                 $index = $name->getAttribute( 'index' );
5543                 return array( $name, $index, $values->item( 0 ) );
5544         }
5545
5546         /**
5547          * Split an <ext> node into an associative array containing name, attr, inner and close
5548          * All values in the resulting array are DOMNodes. Inner and close are optional.
5549          */
5550         function splitExtNode( $node ) {
5551                 $xpath = new DOMXPath( $node->ownerDocument );
5552                 $names = $xpath->query( 'name', $node );
5553                 $attrs = $xpath->query( 'attr', $node );
5554                 $inners = $xpath->query( 'inner', $node );
5555                 $closes = $xpath->query( 'close', $node );
5556                 if ( !$names->length || !$attrs->length ) {
5557                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
5558                 }
5559                 $parts = array(
5560                         'name' => $names->item( 0 ),
5561                         'attr' => $attrs->item( 0 ) );
5562                 if ( $inners->length ) {
5563                         $parts['inner'] = $inners->item( 0 );
5564                 }
5565                 if ( $closes->length ) {
5566                         $parts['close'] = $closes->item( 0 );
5567                 }
5568                 return $parts;
5569         }
5570
5571         function __toString() {
5572                 return 'frame{}';
5573         }
5574
5575         function getPDBK( $level = false ) {
5576                 if ( $level === false ) {
5577                         return $this->title->getPrefixedDBkey();
5578                 } else {
5579                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
5580                 }
5581         }
5582 }
5583
5584 /**
5585  * Expansion frame with template arguments
5586  */
5587 class PPTemplateFrame extends PPFrame {
5588         var $parser, $args, $parent;
5589         var $titleCache;
5590
5591         function __construct( $parser, $parent = false, $args = array(), $title = false ) {
5592                 $this->parser = $parser;
5593                 $this->parent = $parent;
5594                 $this->args = $args;
5595                 $this->title = $title;
5596                 $this->titleCache = $parent->titleCache;
5597                 $this->titleCache[] = $title ? $title->getPrefixedDBkey() : false;
5598         }
5599
5600         function __toString() {
5601                 $s = 'tplframe{';
5602                 $first = true;
5603                 foreach ( $this->args as $name => $value ) {
5604                         if ( $first ) {
5605                                 $first = false;
5606                         } else {
5607                                 $s .= ', ';
5608                         }
5609                         $s .= "\"$name\":\"" .
5610                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5611                 }
5612                 $s .= '}';
5613                 return $s;
5614         }
5615 }
5616