includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10 /**
  11  * Update this version number when the ParserOutput format
  12  * changes in an incompatible way, so the parser cache
  13  * can automatically discard old data.
  14  */
  15 define( 'MW_PARSER_VERSION', '1.6.2' );
  16
  17 define( 'RLH_FOR_UPDATE', 1 );
  18
  19 # Allowed values for $mOutputType
  20 define( 'OT_HTML', 1 );
  21 define( 'OT_WIKI', 2 );
  22 define( 'OT_MSG' , 3 );
  23 define( 'OT_PREPROCESS', 4 );
  24
  25 # Flags for setFunctionHook
  26 define( 'SFH_NO_HASH', 1 );
  27
  28 # string parameter for extractTags which will cause it
  29 # to strip HTML comments in addition to regular
  30 # <XML>-style tags. This should not be anything we
  31 # may want to use in wikisyntax
  32 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  33
  34 # Constants needed for external link processing
  35 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  36 # Everything except bracket, space, or control characters
  37 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  38 # Including space, but excluding newlines
  39 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  40 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  41 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  42 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  43         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  44 define( 'EXT_IMAGE_REGEX',
  45         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  46         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  47         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  48 );
  49
  50 // State constants for the definition list colon extraction
  51 define( 'MW_COLON_STATE_TEXT', 0 );
  52 define( 'MW_COLON_STATE_TAG', 1 );
  53 define( 'MW_COLON_STATE_TAGSTART', 2 );
  54 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  55 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  56 define( 'MW_COLON_STATE_COMMENT', 5 );
  57 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  58 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  59
  60 /**
  61  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  62  * syntax, such as "[[link]]" for making links), and provides a one-way
  63  * transformation of that wiki markup it into XHTML output / markup
  64  * (which in turn the browser understands, and can display).
  65  *
  66  * <pre>
  67  * There are four main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  * preprocess()
  75  *   removes HTML comments and expands templates
  76  *
  77  * Globals used:
  78  *    objects:   $wgLang, $wgContLang
  79  *
  80  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  81  *
  82  * settings:
  83  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  84  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  85  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  86  *  $wgMaxArticleSize*
  87  *
  88  *  * only within ParserOptions
  89  * </pre>
  90  *
  91  * @addtogroup Parser
  92  */
  93 class Parser
  94 {
  95         const VERSION = MW_PARSER_VERSION;
  96         /**#@+
  97          * @private
  98          */
  99         # Persistent:
 100         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
 101                 $mImageParams, $mImageParamsMagicArray;
 102
 103         # Cleared with clearState():
 104         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 105         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 106         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 107         var $mIncludeSizes, $mDefaultSort;
 108         var $mTemplates,        // cache of already loaded templates, avoids
 109                                 // multiple SQL queries for the same string
 110             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 111                                 // in this path. Used for loop detection.
 112
 113         # Temporary
 114         # These are variables reset at least once per parse regardless of $clearState
 115         var $mOptions,      // ParserOptions object
 116                 $mTitle,        // Title context, used for self-link rendering and similar things
 117                 $mOutputType,   // Output type, one of the OT_xxx constants
 118                 $ot,            // Shortcut alias, see setOutputType()
 119                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 120                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 121                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 122
 123         /**#@-*/
 124
 125         /**
 126          * Constructor
 127          *
 128          * @public
 129          */
 130         function Parser() {
 131                 $this->mTagHooks = array();
 132                 $this->mTransparentTagHooks = array();
 133                 $this->mFunctionHooks = array();
 134                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 135                 $this->mFirstCall = true;
 136         }
 137
 138         /**
 139          * Do various kinds of initialisation on the first call of the parser
 140          */
 141         function firstCallInit() {
 142                 if ( !$this->mFirstCall ) {
 143                         return;
 144                 }
 145
 146                 wfProfileIn( __METHOD__ );
 147                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 148
 149                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 150
 151                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 170                 $this->setFunctionHook( 'numberofedits', array( 'CoreParserFunctions', 'numberofedits' ), SFH_NO_HASH );
 171                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 172                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 173                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 174                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 175                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 176                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 177                 $this->setFunctionHook( 'filepath', array( 'CoreParserFunctions', 'filepath' ), SFH_NO_HASH );
 178
 179                 if ( $wgAllowDisplayTitle ) {
 180                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 181                 }
 182                 if ( $wgAllowSlowParserFunctions ) {
 183                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 184                 }
 185
 186                 $this->initialiseVariables();
 187                 $this->mFirstCall = false;
 188                 wfProfileOut( __METHOD__ );
 189         }
 190
 191         /**
 192          * Clear Parser state
 193          *
 194          * @private
 195          */
 196         function clearState() {
 197                 wfProfileIn( __METHOD__ );
 198                 if ( $this->mFirstCall ) {
 199                         $this->firstCallInit();
 200                 }
 201                 $this->mOutput = new ParserOutput;
 202                 $this->mAutonumber = 0;
 203                 $this->mLastSection = '';
 204                 $this->mDTopen = false;
 205                 $this->mIncludeCount = array();
 206                 $this->mStripState = new StripState;
 207                 $this->mArgStack = array();
 208                 $this->mInPre = false;
 209                 $this->mInterwikiLinkHolders = array(
 210                         'texts' => array(),
 211                         'titles' => array()
 212                 );
 213                 $this->mLinkHolders = array(
 214                         'namespaces' => array(),
 215                         'dbkeys' => array(),
 216                         'queries' => array(),
 217                         'texts' => array(),
 218                         'titles' => array()
 219                 );
 220                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 221
 222                 /**
 223                  * Prefix for temporary replacement strings for the multipass parser.
 224                  * \x07 should never appear in input as it's disallowed in XML.
 225                  * Using it at the front also gives us a little extra robustness
 226                  * since it shouldn't match when butted up against identifier-like
 227                  * string constructs.
 228                  */
 229                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 230
 231                 # Clear these on every parse, bug 4549
 232                 $this->mTemplates = array();
 233                 $this->mTemplatePath = array();
 234
 235                 $this->mShowToc = true;
 236                 $this->mForceTocPosition = false;
 237                 $this->mIncludeSizes = array(
 238                         'pre-expand' => 0,
 239                         'post-expand' => 0,
 240                         'arg' => 0
 241                 );
 242                 $this->mDefaultSort = false;
 243
 244                 wfRunHooks( 'ParserClearState', array( &$this ) );
 245                 wfProfileOut( __METHOD__ );
 246         }
 247
 248         function setOutputType( $ot ) {
 249                 $this->mOutputType = $ot;
 250                 // Shortcut alias
 251                 $this->ot = array(
 252                         'html' => $ot == OT_HTML,
 253                         'wiki' => $ot == OT_WIKI,
 254                         'msg' => $ot == OT_MSG,
 255                         'pre' => $ot == OT_PREPROCESS,
 256                 );
 257         }
 258
 259         /**
 260          * Accessor for mUniqPrefix.
 261          *
 262          * @public
 263          */
 264         function uniqPrefix() {
 265                 return $this->mUniqPrefix;
 266         }
 267
 268         /**
 269          * Convert wikitext to HTML
 270          * Do not call this function recursively.
 271          *
 272          * @param string $text Text we want to parse
 273          * @param Title &$title A title object
 274          * @param array $options
 275          * @param boolean $linestart
 276          * @param boolean $clearState
 277          * @param int $revid number to pass in {{REVISIONID}}
 278          * @return ParserOutput a ParserOutput
 279          */
 280         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 281                 /**
 282                  * First pass--just handle <nowiki> sections, pass the rest off
 283                  * to internalParse() which does all the real work.
 284                  */
 285
 286                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 287                 $fname = 'Parser::parse-' . wfGetCaller();
 288                 wfProfileIn( __METHOD__ );
 289                 wfProfileIn( $fname );
 290
 291                 if ( $clearState ) {
 292                         $this->clearState();
 293                 }
 294
 295                 $this->mOptions = $options;
 296                 $this->mTitle =& $title;
 297                 $oldRevisionId = $this->mRevisionId;
 298                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 299                 if( $revid !== null ) {
 300                         $this->mRevisionId = $revid;
 301                         $this->mRevisionTimestamp = null;
 302                 }
 303                 $this->setOutputType( OT_HTML );
 304                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 305                 $text = $this->strip( $text, $this->mStripState );
 306                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 307                 $text = $this->internalParse( $text );
 308                 $text = $this->mStripState->unstripGeneral( $text );
 309
 310                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 311                 $fixtags = array(
 312                         # french spaces, last one Guillemet-left
 313                         # only if there is something before the space
 314                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 315                         # french spaces, Guillemet-right
 316                         '/(\\302\\253) /' => '\\1&nbsp;',
 317                 );
 318                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 319
 320                 # only once and last
 321                 $text = $this->doBlockLevels( $text, $linestart );
 322
 323                 $this->replaceLinkHolders( $text );
 324
 325                 # the position of the parserConvert() call should not be changed. it
 326                 # assumes that the links are all replaced and the only thing left
 327                 # is the <nowiki> mark.
 328                 # Side-effects: this calls $this->mOutput->setTitleText()
 329                 $text = $wgContLang->parserConvert( $text, $this );
 330
 331                 $text = $this->mStripState->unstripNoWiki( $text );
 332
 333                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 334
 335 //!JF Move to its own function
 336
 337                 $uniq_prefix = $this->mUniqPrefix;
 338                 $matches = array();
 339                 $elements = array_keys( $this->mTransparentTagHooks );
 340                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 341
 342                 foreach( $matches as $marker => $data ) {
 343                         list( $element, $content, $params, $tag ) = $data;
 344                         $tagName = strtolower( $element );
 345                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 346                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 347                                         array( $content, $params, $this ) );
 348                         } else {
 349                                 $output = $tag;
 350                         }
 351                         $this->mStripState->general->setPair( $marker, $output );
 352                 }
 353                 $text = $this->mStripState->unstripGeneral( $text );
 354
 355                 $text = Sanitizer::normalizeCharReferences( $text );
 356
 357                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 358                         $text = Parser::tidy($text);
 359                 } else {
 360                         # attempt to sanitize at least some nesting problems
 361                         # (bug #2702 and quite a few others)
 362                         $tidyregs = array(
 363                                 # ''Something [http://www.cool.com cool''] -->
 364                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 365                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 366                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 367                                 # fix up an anchor inside another anchor, only
 368                                 # at least for a single single nested link (bug 3695)
 369                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 370                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 371                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 372                                 # contains a div, so fix it up here; replace
 373                                 # div with escaped text
 374                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 375                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 376                                 # remove empty italic or bold tag pairs, some
 377                                 # introduced by rules above
 378                                 '/<([bi])><\/\\1>/' => '',
 379                         );
 380
 381                         $text = preg_replace(
 382                                 array_keys( $tidyregs ),
 383                                 array_values( $tidyregs ),
 384                                 $text );
 385                 }
 386
 387                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 388
 389                 # Information on include size limits, for the benefit of users who try to skirt them
 390                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 391                         $max = $this->mOptions->getMaxIncludeSize();
 392                         $text .= "<!-- \n" .
 393                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 394                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 395                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 396                                 "Maximum: $max bytes\n" .
 397                                 "-->\n";
 398                 }
 399                 $this->mOutput->setText( $text );
 400                 $this->mRevisionId = $oldRevisionId;
 401                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 402                 wfProfileOut( $fname );
 403                 wfProfileOut( __METHOD__ );
 404
 405                 return $this->mOutput;
 406         }
 407
 408         /**
 409          * Recursive parser entry point that can be called from an extension tag
 410          * hook.
 411          */
 412         function recursiveTagParse( $text ) {
 413                 wfProfileIn( __METHOD__ );
 414                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 415                 $text = $this->strip( $text, $this->mStripState );
 416                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 417                 $text = $this->internalParse( $text );
 418                 wfProfileOut( __METHOD__ );
 419                 return $text;
 420         }
 421
 422         /**
 423          * Expand templates and variables in the text, producing valid, static wikitext.
 424          * Also removes comments.
 425          */
 426         function preprocess( $text, $title, $options, $revid = null ) {
 427                 wfProfileIn( __METHOD__ );
 428                 $this->clearState();
 429                 $this->setOutputType( OT_PREPROCESS );
 430                 $this->mOptions = $options;
 431                 $this->mTitle = $title;
 432                 if( $revid !== null ) {
 433                         $this->mRevisionId = $revid;
 434                 }
 435                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 436                 $text = $this->strip( $text, $this->mStripState );
 437                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 438                 if ( $this->mOptions->getRemoveComments() ) {
 439                         $text = Sanitizer::removeHTMLcomments( $text );
 440                 }
 441                 $text = $this->replaceVariables( $text );
 442                 $text = $this->mStripState->unstripBoth( $text );
 443                 wfProfileOut( __METHOD__ );
 444                 return $text;
 445         }
 446
 447         /**
 448          * Get a random string
 449          *
 450          * @private
 451          * @static
 452          */
 453         function getRandomString() {
 454                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 455         }
 456
 457         function &getTitle() { return $this->mTitle; }
 458         function getOptions() { return $this->mOptions; }
 459
 460         function getFunctionLang() {
 461                 global $wgLang, $wgContLang;
 462                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 463         }
 464
 465         /**
 466          * Replaces all occurrences of HTML-style comments and the given tags
 467          * in the text with a random marker and returns teh next text. The output
 468          * parameter $matches will be an associative array filled with data in
 469          * the form:
 470          *   'UNIQ-xxxxx' => array(
 471          *     'element',
 472          *     'tag content',
 473          *     array( 'param' => 'x' ),
 474          *     '<element param="x">tag content</element>' ) )
 475          *
 476          * @param $elements list of element names. Comments are always extracted.
 477          * @param $text Source text string.
 478          * @param $uniq_prefix
 479          *
 480          * @public
 481          * @static
 482          */
 483         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 484                 static $n = 1;
 485                 $stripped = '';
 486                 $matches = array();
 487
 488                 $taglist = implode( '|', $elements );
 489                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 490
 491                 while ( '' != $text ) {
 492                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 493                         $stripped .= $p[0];
 494                         if( count( $p ) < 5 ) {
 495                                 break;
 496                         }
 497                         if( count( $p ) > 5 ) {
 498                                 // comment
 499                                 $element    = $p[4];
 500                                 $attributes = '';
 501                                 $close      = '';
 502                                 $inside     = $p[5];
 503                         } else {
 504                                 // tag
 505                                 $element    = $p[1];
 506                                 $attributes = $p[2];
 507                                 $close      = $p[3];
 508                                 $inside     = $p[4];
 509                         }
 510
 511                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 512                         $stripped .= $marker;
 513
 514                         if ( $close === '/>' ) {
 515                                 // Empty element tag, <tag />
 516                                 $content = null;
 517                                 $text = $inside;
 518                                 $tail = null;
 519                         } else {
 520                                 if( $element == '!--' ) {
 521                                         $end = '/(-->)/';
 522                                 } else {
 523                                         $end = "/(<\\/$element\\s*>)/i";
 524                                 }
 525                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 526                                 $content = $q[0];
 527                                 if( count( $q ) < 3 ) {
 528                                         # No end tag -- let it run out to the end of the text.
 529                                         $tail = '';
 530                                         $text = '';
 531                                 } else {
 532                                         $tail = $q[1];
 533                                         $text = $q[2];
 534                                 }
 535                         }
 536
 537                         $matches[$marker] = array( $element,
 538                                 $content,
 539                                 Sanitizer::decodeTagAttributes( $attributes ),
 540                                 "<$element$attributes$close$content$tail" );
 541                 }
 542                 return $stripped;
 543         }
 544
 545         /**
 546          * Strips and renders nowiki, pre, math, hiero
 547          * If $render is set, performs necessary rendering operations on plugins
 548          * Returns the text, and fills an array with data needed in unstrip()
 549          *
 550          * @param StripState $state
 551          *
 552          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 553          *  will be stripped in addition to other tags. This is important
 554          *  for section editing, where these comments cause confusion when
 555          *  counting the sections in the wikisource
 556          *
 557          * @param array dontstrip contains tags which should not be stripped;
 558          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 559          *
 560          * @private
 561          */
 562         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 563                 global $wgContLang;
 564                 wfProfileIn( __METHOD__ );
 565                 $render = ($this->mOutputType == OT_HTML);
 566
 567                 $uniq_prefix = $this->mUniqPrefix;
 568                 $commentState = new ReplacementArray;
 569                 $nowikiItems = array();
 570                 $generalItems = array();
 571
 572                 $elements = array_merge(
 573                         array( 'nowiki', 'gallery' ),
 574                         array_keys( $this->mTagHooks ) );
 575                 global $wgRawHtml;
 576                 if( $wgRawHtml ) {
 577                         $elements[] = 'html';
 578                 }
 579                 if( $this->mOptions->getUseTeX() ) {
 580                         $elements[] = 'math';
 581                 }
 582
 583                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 584                 foreach ( $elements AS $k => $v ) {
 585                         if ( !in_array ( $v , $dontstrip ) ) continue;
 586                         unset ( $elements[$k] );
 587                 }
 588
 589                 $matches = array();
 590                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 591
 592                 foreach( $matches as $marker => $data ) {
 593                         list( $element, $content, $params, $tag ) = $data;
 594                         if( $render ) {
 595                                 $tagName = strtolower( $element );
 596                                 wfProfileIn( __METHOD__."-render-$tagName" );
 597                                 switch( $tagName ) {
 598                                 case '!--':
 599                                         // Comment
 600                                         if( substr( $tag, -3 ) == '-->' ) {
 601                                                 $output = $tag;
 602                                         } else {
 603                                                 // Unclosed comment in input.
 604                                                 // Close it so later stripping can remove it
 605                                                 $output = "$tag-->";
 606                                         }
 607                                         break;
 608                                 case 'html':
 609                                         if( $wgRawHtml ) {
 610                                                 $output = $content;
 611                                                 break;
 612                                         }
 613                                         // Shouldn't happen otherwise. :)
 614                                 case 'nowiki':
 615                                         $output = Xml::escapeTagsOnly( $content );
 616                                         break;
 617                                 case 'math':
 618                                         $output = $wgContLang->armourMath(
 619                                                 MathRenderer::renderMath( $content, $params ) );
 620                                         break;
 621                                 case 'gallery':
 622                                         $output = $this->renderImageGallery( $content, $params );
 623                                         break;
 624                                 default:
 625                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 626                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 627                                                         array( $content, $params, $this ) );
 628                                         } else {
 629                                                 throw new MWException( "Invalid call hook $element" );
 630                                         }
 631                                 }
 632                                 wfProfileOut( __METHOD__."-render-$tagName" );
 633                         } else {
 634                                 // Just stripping tags; keep the source
 635                                 $output = $tag;
 636                         }
 637
 638                         // Unstrip the output, to support recursive strip() calls
 639                         $output = $state->unstripBoth( $output );
 640
 641                         if( !$stripcomments && $element == '!--' ) {
 642                                 $commentState->setPair( $marker, $output );
 643                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 644                                 $nowikiItems[$marker] = $output;
 645                         } else {
 646                                 $generalItems[$marker] = $output;
 647                         }
 648                 }
 649                 # Add the new items to the state
 650                 # We do this after the loop instead of during it to avoid slowing
 651                 # down the recursive unstrip
 652                 $state->nowiki->mergeArray( $nowikiItems );
 653                 $state->general->mergeArray( $generalItems );
 654
 655                 # Unstrip comments unless explicitly told otherwise.
 656                 # (The comments are always stripped prior to this point, so as to
 657                 # not invoke any extension tags / parser hooks contained within
 658                 # a comment.)
 659                 if ( !$stripcomments ) {
 660                         // Put them all back and forget them
 661                         $text = $commentState->replace( $text );
 662                 }
 663
 664                 wfProfileOut( __METHOD__ );
 665                 return $text;
 666         }
 667
 668         /**
 669          * Restores pre, math, and other extensions removed by strip()
 670          *
 671          * always call unstripNoWiki() after this one
 672          * @private
 673          * @deprecated use $this->mStripState->unstrip()
 674          */
 675         function unstrip( $text, $state ) {
 676                 return $state->unstripGeneral( $text );
 677         }
 678
 679         /**
 680          * Always call this after unstrip() to preserve the order
 681          *
 682          * @private
 683          * @deprecated use $this->mStripState->unstrip()
 684          */
 685         function unstripNoWiki( $text, $state ) {
 686                 return $state->unstripNoWiki( $text );
 687         }
 688
 689         /**
 690          * @deprecated use $this->mStripState->unstripBoth()
 691          */
 692         function unstripForHTML( $text ) {
 693                 return $this->mStripState->unstripBoth( $text );
 694         }
 695
 696         /**
 697          * Add an item to the strip state
 698          * Returns the unique tag which must be inserted into the stripped text
 699          * The tag will be replaced with the original text in unstrip()
 700          *
 701          * @private
 702          */
 703         function insertStripItem( $text, &$state ) {
 704                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 705                 $state->general->setPair( $rnd, $text );
 706                 return $rnd;
 707         }
 708
 709         /**
 710          * Interface with html tidy, used if $wgUseTidy = true.
 711          * If tidy isn't able to correct the markup, the original will be
 712          * returned in all its glory with a warning comment appended.
 713          *
 714          * Either the external tidy program or the in-process tidy extension
 715          * will be used depending on availability. Override the default
 716          * $wgTidyInternal setting to disable the internal if it's not working.
 717          *
 718          * @param string $text Hideous HTML input
 719          * @return string Corrected HTML output
 720          * @public
 721          * @static
 722          */
 723         function tidy( $text ) {
 724                 global $wgTidyInternal;
 725                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 726 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 727 '<head><title>test</title></head><body>'.$text.'</body></html>';
 728                 if( $wgTidyInternal ) {
 729                         $correctedtext = Parser::internalTidy( $wrappedtext );
 730                 } else {
 731                         $correctedtext = Parser::externalTidy( $wrappedtext );
 732                 }
 733                 if( is_null( $correctedtext ) ) {
 734                         wfDebug( "Tidy error detected!\n" );
 735                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 736                 }
 737                 return $correctedtext;
 738         }
 739
 740         /**
 741          * Spawn an external HTML tidy process and get corrected markup back from it.
 742          *
 743          * @private
 744          * @static
 745          */
 746         function externalTidy( $text ) {
 747                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 748                 $fname = 'Parser::externalTidy';
 749                 wfProfileIn( $fname );
 750
 751                 $cleansource = '';
 752                 $opts = ' -utf8';
 753
 754                 $descriptorspec = array(
 755                         0 => array('pipe', 'r'),
 756                         1 => array('pipe', 'w'),
 757                         2 => array('file', wfGetNull(), 'a')
 758                 );
 759                 $pipes = array();
 760                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 761                 if (is_resource($process)) {
 762                         // Theoretically, this style of communication could cause a deadlock
 763                         // here. If the stdout buffer fills up, then writes to stdin could
 764                         // block. This doesn't appear to happen with tidy, because tidy only
 765                         // writes to stdout after it's finished reading from stdin. Search
 766                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 767                         fwrite($pipes[0], $text);
 768                         fclose($pipes[0]);
 769                         while (!feof($pipes[1])) {
 770                                 $cleansource .= fgets($pipes[1], 1024);
 771                         }
 772                         fclose($pipes[1]);
 773                         proc_close($process);
 774                 }
 775
 776                 wfProfileOut( $fname );
 777
 778                 if( $cleansource == '' && $text != '') {
 779                         // Some kind of error happened, so we couldn't get the corrected text.
 780                         // Just give up; we'll use the source text and append a warning.
 781                         return null;
 782                 } else {
 783                         return $cleansource;
 784                 }
 785         }
 786
 787         /**
 788          * Use the HTML tidy PECL extension to use the tidy library in-process,
 789          * saving the overhead of spawning a new process. Currently written to
 790          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 791          *
 792          * 'pear install tidy' should be able to compile the extension module.
 793          *
 794          * @private
 795          * @static
 796          */
 797         function internalTidy( $text ) {
 798                 global $wgTidyConf;
 799                 $fname = 'Parser::internalTidy';
 800                 wfProfileIn( $fname );
 801
 802                 tidy_load_config( $wgTidyConf );
 803                 tidy_set_encoding( 'utf8' );
 804                 tidy_parse_string( $text );
 805                 tidy_clean_repair();
 806                 if( tidy_get_status() == 2 ) {
 807                         // 2 is magic number for fatal error
 808                         // http://www.php.net/manual/en/function.tidy-get-status.php
 809                         $cleansource = null;
 810                 } else {
 811                         $cleansource = tidy_get_output();
 812                 }
 813                 wfProfileOut( $fname );
 814                 return $cleansource;
 815         }
 816
 817         /**
 818          * parse the wiki syntax used to render tables
 819          *
 820          * @private
 821          */
 822         function doTableStuff ( $text ) {
 823                 $fname = 'Parser::doTableStuff';
 824                 wfProfileIn( $fname );
 825
 826                 $lines = explode ( "\n" , $text );
 827                 $td_history = array (); // Is currently a td tag open?
 828                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 829                 $tr_history = array (); // Is currently a tr tag open?
 830                 $tr_attributes = array (); // history of tr attributes
 831                 $has_opened_tr = array(); // Did this table open a <tr> element?
 832                 $indent_level = 0; // indent level of the table
 833                 foreach ( $lines as $key => $line )
 834                 {
 835                         $line = trim ( $line );
 836
 837                         if( $line == '' ) { // empty line, go to next line
 838                                 continue;
 839                         }
 840                         $first_character = $line{0};
 841                         $matches = array();
 842
 843                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 844                                 // First check if we are starting a new table
 845                                 $indent_level = strlen( $matches[1] );
 846
 847                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 848                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 849
 850                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 851                                 array_push ( $td_history , false );
 852                                 array_push ( $last_tag_history , '' );
 853                                 array_push ( $tr_history , false );
 854                                 array_push ( $tr_attributes , '' );
 855                                 array_push ( $has_opened_tr , false );
 856                         } else if ( count ( $td_history ) == 0 ) {
 857                                 // Don't do any of the following
 858                                 continue;
 859                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 860                                 // We are ending a table
 861                                 $line = '</table>' . substr ( $line , 2 );
 862                                 $last_tag = array_pop ( $last_tag_history );
 863
 864                                 if ( !array_pop ( $has_opened_tr ) ) {
 865                                         $line = "<tr><td></td></tr>{$line}";
 866                                 }
 867
 868                                 if ( array_pop ( $tr_history ) ) {
 869                                         $line = "</tr>{$line}";
 870                                 }
 871
 872                                 if ( array_pop ( $td_history ) ) {
 873                                         $line = "</{$last_tag}>{$line}";
 874                                 }
 875                                 array_pop ( $tr_attributes );
 876                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 877                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 878                                 // Now we have a table row
 879                                 $line = preg_replace( '#^\|-+#', '', $line );
 880
 881                                 // Whats after the tag is now only attributes
 882                                 $attributes = $this->mStripState->unstripBoth( $line );
 883                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 884                                 array_pop ( $tr_attributes );
 885                                 array_push ( $tr_attributes , $attributes );
 886
 887                                 $line = '';
 888                                 $last_tag = array_pop ( $last_tag_history );
 889                                 array_pop ( $has_opened_tr );
 890                                 array_push ( $has_opened_tr , true );
 891
 892                                 if ( array_pop ( $tr_history ) ) {
 893                                         $line = '</tr>';
 894                                 }
 895
 896                                 if ( array_pop ( $td_history ) ) {
 897                                         $line = "</{$last_tag}>{$line}";
 898                                 }
 899
 900                                 $lines[$key] = $line;
 901                                 array_push ( $tr_history , false );
 902                                 array_push ( $td_history , false );
 903                                 array_push ( $last_tag_history , '' );
 904                         }
 905                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 906                                 // This might be cell elements, td, th or captions
 907                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 908                                         $first_character = '+';
 909                                         $line = substr ( $line , 1 );
 910                                 }
 911
 912                                 $line = substr ( $line , 1 );
 913
 914                                 if ( $first_character == '!' ) {
 915                                         $line = str_replace ( '!!' , '||' , $line );
 916                                 }
 917
 918                                 // Split up multiple cells on the same line.
 919                                 // FIXME : This can result in improper nesting of tags processed
 920                                 // by earlier parser steps, but should avoid splitting up eg
 921                                 // attribute values containing literal "||".
 922                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 923
 924                                 $lines[$key] = '';
 925
 926                                 // Loop through each table cell
 927                                 foreach ( $cells as $cell )
 928                                 {
 929                                         $previous = '';
 930                                         if ( $first_character != '+' )
 931                                         {
 932                                                 $tr_after = array_pop ( $tr_attributes );
 933                                                 if ( !array_pop ( $tr_history ) ) {
 934                                                         $previous = "<tr{$tr_after}>\n";
 935                                                 }
 936                                                 array_push ( $tr_history , true );
 937                                                 array_push ( $tr_attributes , '' );
 938                                                 array_pop ( $has_opened_tr );
 939                                                 array_push ( $has_opened_tr , true );
 940                                         }
 941
 942                                         $last_tag = array_pop ( $last_tag_history );
 943
 944                                         if ( array_pop ( $td_history ) ) {
 945                                                 $previous = "</{$last_tag}>{$previous}";
 946                                         }
 947
 948                                         if ( $first_character == '|' ) {
 949                                                 $last_tag = 'td';
 950                                         } else if ( $first_character == '!' ) {
 951                                                 $last_tag = 'th';
 952                                         } else if ( $first_character == '+' ) {
 953                                                 $last_tag = 'caption';
 954                                         } else {
 955                                                 $last_tag = '';
 956                                         }
 957
 958                                         array_push ( $last_tag_history , $last_tag );
 959
 960                                         // A cell could contain both parameters and data
 961                                         $cell_data = explode ( '|' , $cell , 2 );
 962
 963                                         // Bug 553: Note that a '|' inside an invalid link should not
 964                                         // be mistaken as delimiting cell parameters
 965                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 966                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 967                                         } else if ( count ( $cell_data ) == 1 )
 968                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 969                                         else {
 970                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 971                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 972                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 973                                         }
 974
 975                                         $lines[$key] .= $cell;
 976                                         array_push ( $td_history , true );
 977                                 }
 978                         }
 979                 }
 980
 981                 // Closing open td, tr && table
 982                 while ( count ( $td_history ) > 0 )
 983                 {
 984                         if ( array_pop ( $td_history ) ) {
 985                                 $lines[] = '</td>' ;
 986                         }
 987                         if ( array_pop ( $tr_history ) ) {
 988                                 $lines[] = '</tr>' ;
 989                         }
 990                         if ( !array_pop ( $has_opened_tr ) ) {
 991                                 $lines[] = "<tr><td></td></tr>" ;
 992                         }
 993
 994                         $lines[] = '</table>' ;
 995                 }
 996
 997                 $output = implode ( "\n" , $lines ) ;
 998
 999                 // special case: don't return empty table
1000                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
1001                         $output = '';
1002                 }
1003
1004                 wfProfileOut( $fname );
1005
1006                 return $output;
1007         }
1008
1009         /**
1010          * Helper function for parse() that transforms wiki markup into
1011          * HTML. Only called for $mOutputType == OT_HTML.
1012          *
1013          * @private
1014          */
1015         function internalParse( $text ) {
1016                 $args = array();
1017                 $isMain = true;
1018                 $fname = 'Parser::internalParse';
1019                 wfProfileIn( $fname );
1020
1021                 # Hook to suspend the parser in this state
1022                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1023                         wfProfileOut( $fname );
1024                         return $text ;
1025                 }
1026
1027                 # Remove <noinclude> tags and <includeonly> sections
1028                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
1029                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
1030                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
1031
1032                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
1033
1034                 $text = $this->replaceVariables( $text, $args );
1035                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1036
1037                 // Tables need to come after variable replacement for things to work
1038                 // properly; putting them before other transformations should keep
1039                 // exciting things like link expansions from showing up in surprising
1040                 // places.
1041                 $text = $this->doTableStuff( $text );
1042
1043                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1044
1045                 $text = $this->stripToc( $text );
1046                 $this->stripNoGallery( $text );
1047                 $text = $this->doHeadings( $text );
1048                 if($this->mOptions->getUseDynamicDates()) {
1049                         $df =& DateFormatter::getInstance();
1050                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1051                 }
1052                 $text = $this->doAllQuotes( $text );
1053                 $text = $this->replaceInternalLinks( $text );
1054                 $text = $this->replaceExternalLinks( $text );
1055
1056                 # replaceInternalLinks may sometimes leave behind
1057                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1058                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1059
1060                 $text = $this->doMagicLinks( $text );
1061                 $text = $this->formatHeadings( $text, $isMain );
1062
1063                 wfProfileOut( $fname );
1064                 return $text;
1065         }
1066
1067         /**
1068          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1069          * magic external links.
1070          *
1071          * @private
1072          */
1073         function &doMagicLinks( &$text ) {
1074                 wfProfileIn( __METHOD__ );
1075                 $text = preg_replace_callback(
1076                         '!(?:                           # Start cases
1077                             <a.*?</a> |                 # Skip link text
1078                             <.*?> |                     # Skip stuff inside HTML elements
1079                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1080                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1081                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1082                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1083                                      [0-9Xx]                 # check digit
1084                                    \b)
1085                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1086                 wfProfileOut( __METHOD__ );
1087                 return $text;
1088         }
1089
1090         function magicLinkCallback( $m ) {
1091                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1092                         # Skip HTML element
1093                         return $m[0];
1094                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1095                         $isbn = $m[2];
1096                         $num = strtr( $isbn, array(
1097                                 '-' => '',
1098                                 ' ' => '',
1099                                 'x' => 'X',
1100                         ));
1101                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1102                         $text = '<a href="' .
1103                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1104                                 "\" class=\"internal\">ISBN $isbn</a>";
1105                 } else {
1106                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1107                                 $keyword = 'RFC';
1108                                 $urlmsg = 'rfcurl';
1109                                 $id = $m[1];
1110                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1111                                 $keyword = 'PMID';
1112                                 $urlmsg = 'pubmedurl';
1113                                 $id = $m[1];
1114                         } else {
1115                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1116                                         substr($m[0], 0, 20 ) . '"' );
1117                         }
1118
1119                         $url = wfMsg( $urlmsg, $id);
1120                         $sk = $this->mOptions->getSkin();
1121                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1122                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1123                 }
1124                 return $text;
1125         }
1126
1127         /**
1128          * Parse headers and return html
1129          *
1130          * @private
1131          */
1132         function doHeadings( $text ) {
1133                 $fname = 'Parser::doHeadings';
1134                 wfProfileIn( $fname );
1135                 for ( $i = 6; $i >= 1; --$i ) {
1136                         $h = str_repeat( '=', $i );
1137                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1138                           "<h{$i}>\\1</h{$i}>\\2", $text );
1139                 }
1140                 wfProfileOut( $fname );
1141                 return $text;
1142         }
1143
1144         /**
1145          * Replace single quotes with HTML markup
1146          * @private
1147          * @return string the altered text
1148          */
1149         function doAllQuotes( $text ) {
1150                 $fname = 'Parser::doAllQuotes';
1151                 wfProfileIn( $fname );
1152                 $outtext = '';
1153                 $lines = explode( "\n", $text );
1154                 foreach ( $lines as $line ) {
1155                         $outtext .= $this->doQuotes ( $line ) . "\n";
1156                 }
1157                 $outtext = substr($outtext, 0,-1);
1158                 wfProfileOut( $fname );
1159                 return $outtext;
1160         }
1161
1162         /**
1163          * Helper function for doAllQuotes()
1164          */
1165         public function doQuotes( $text ) {
1166                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1167                 if ( count( $arr ) == 1 )
1168                         return $text;
1169                 else
1170                 {
1171                         # First, do some preliminary work. This may shift some apostrophes from
1172                         # being mark-up to being text. It also counts the number of occurrences
1173                         # of bold and italics mark-ups.
1174                         $i = 0;
1175                         $numbold = 0;
1176                         $numitalics = 0;
1177                         foreach ( $arr as $r )
1178                         {
1179                                 if ( ( $i % 2 ) == 1 )
1180                                 {
1181                                         # If there are ever four apostrophes, assume the first is supposed to
1182                                         # be text, and the remaining three constitute mark-up for bold text.
1183                                         if ( strlen( $arr[$i] ) == 4 )
1184                                         {
1185                                                 $arr[$i-1] .= "'";
1186                                                 $arr[$i] = "'''";
1187                                         }
1188                                         # If there are more than 5 apostrophes in a row, assume they're all
1189                                         # text except for the last 5.
1190                                         else if ( strlen( $arr[$i] ) > 5 )
1191                                         {
1192                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1193                                                 $arr[$i] = "'''''";
1194                                         }
1195                                         # Count the number of occurrences of bold and italics mark-ups.
1196                                         # We are not counting sequences of five apostrophes.
1197                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1198                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1199                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1200                                 }
1201                                 $i++;
1202                         }
1203
1204                         # If there is an odd number of both bold and italics, it is likely
1205                         # that one of the bold ones was meant to be an apostrophe followed
1206                         # by italics. Which one we cannot know for certain, but it is more
1207                         # likely to be one that has a single-letter word before it.
1208                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1209                         {
1210                                 $i = 0;
1211                                 $firstsingleletterword = -1;
1212                                 $firstmultiletterword = -1;
1213                                 $firstspace = -1;
1214                                 foreach ( $arr as $r )
1215                                 {
1216                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1217                                         {
1218                                                 $x1 = substr ($arr[$i-1], -1);
1219                                                 $x2 = substr ($arr[$i-1], -2, 1);
1220                                                 if ($x1 == ' ') {
1221                                                         if ($firstspace == -1) $firstspace = $i;
1222                                                 } else if ($x2 == ' ') {
1223                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1224                                                 } else {
1225                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1226                                                 }
1227                                         }
1228                                         $i++;
1229                                 }
1230
1231                                 # If there is a single-letter word, use it!
1232                                 if ($firstsingleletterword > -1)
1233                                 {
1234                                         $arr [ $firstsingleletterword ] = "''";
1235                                         $arr [ $firstsingleletterword-1 ] .= "'";
1236                                 }
1237                                 # If not, but there's a multi-letter word, use that one.
1238                                 else if ($firstmultiletterword > -1)
1239                                 {
1240                                         $arr [ $firstmultiletterword ] = "''";
1241                                         $arr [ $firstmultiletterword-1 ] .= "'";
1242                                 }
1243                                 # ... otherwise use the first one that has neither.
1244                                 # (notice that it is possible for all three to be -1 if, for example,
1245                                 # there is only one pentuple-apostrophe in the line)
1246                                 else if ($firstspace > -1)
1247                                 {
1248                                         $arr [ $firstspace ] = "''";
1249                                         $arr [ $firstspace-1 ] .= "'";
1250                                 }
1251                         }
1252
1253                         # Now let's actually convert our apostrophic mush to HTML!
1254                         $output = '';
1255                         $buffer = '';
1256                         $state = '';
1257                         $i = 0;
1258                         foreach ($arr as $r)
1259                         {
1260                                 if (($i % 2) == 0)
1261                                 {
1262                                         if ($state == 'both')
1263                                                 $buffer .= $r;
1264                                         else
1265                                                 $output .= $r;
1266                                 }
1267                                 else
1268                                 {
1269                                         if (strlen ($r) == 2)
1270                                         {
1271                                                 if ($state == 'i')
1272                                                 { $output .= '</i>'; $state = ''; }
1273                                                 else if ($state == 'bi')
1274                                                 { $output .= '</i>'; $state = 'b'; }
1275                                                 else if ($state == 'ib')
1276                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1277                                                 else if ($state == 'both')
1278                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1279                                                 else # $state can be 'b' or ''
1280                                                 { $output .= '<i>'; $state .= 'i'; }
1281                                         }
1282                                         else if (strlen ($r) == 3)
1283                                         {
1284                                                 if ($state == 'b')
1285                                                 { $output .= '</b>'; $state = ''; }
1286                                                 else if ($state == 'bi')
1287                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1288                                                 else if ($state == 'ib')
1289                                                 { $output .= '</b>'; $state = 'i'; }
1290                                                 else if ($state == 'both')
1291                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1292                                                 else # $state can be 'i' or ''
1293                                                 { $output .= '<b>'; $state .= 'b'; }
1294                                         }
1295                                         else if (strlen ($r) == 5)
1296                                         {
1297                                                 if ($state == 'b')
1298                                                 { $output .= '</b><i>'; $state = 'i'; }
1299                                                 else if ($state == 'i')
1300                                                 { $output .= '</i><b>'; $state = 'b'; }
1301                                                 else if ($state == 'bi')
1302                                                 { $output .= '</i></b>'; $state = ''; }
1303                                                 else if ($state == 'ib')
1304                                                 { $output .= '</b></i>'; $state = ''; }
1305                                                 else if ($state == 'both')
1306                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1307                                                 else # ($state == '')
1308                                                 { $buffer = ''; $state = 'both'; }
1309                                         }
1310                                 }
1311                                 $i++;
1312                         }
1313                         # Now close all remaining tags.  Notice that the order is important.
1314                         if ($state == 'b' || $state == 'ib')
1315                                 $output .= '</b>';
1316                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1317                                 $output .= '</i>';
1318                         if ($state == 'bi')
1319                                 $output .= '</b>';
1320                         # There might be lonely ''''', so make sure we have a buffer
1321                         if ($state == 'both' && $buffer)
1322                                 $output .= '<b><i>'.$buffer.'</i></b>';
1323                         return $output;
1324                 }
1325         }
1326
1327         /**
1328          * Replace external links
1329          *
1330          * Note: this is all very hackish and the order of execution matters a lot.
1331          * Make sure to run maintenance/parserTests.php if you change this code.
1332          *
1333          * @private
1334          */
1335         function replaceExternalLinks( $text ) {
1336                 global $wgContLang;
1337                 $fname = 'Parser::replaceExternalLinks';
1338                 wfProfileIn( $fname );
1339
1340                 $sk = $this->mOptions->getSkin();
1341
1342                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1343
1344                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1345
1346                 $i = 0;
1347                 while ( $i<count( $bits ) ) {
1348                         $url = $bits[$i++];
1349                         $protocol = $bits[$i++];
1350                         $text = $bits[$i++];
1351                         $trail = $bits[$i++];
1352
1353                         # The characters '<' and '>' (which were escaped by
1354                         # removeHTMLtags()) should not be included in
1355                         # URLs, per RFC 2396.
1356                         $m2 = array();
1357                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1358                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1359                                 $url = substr($url, 0, $m2[0][1]);
1360                         }
1361
1362                         # If the link text is an image URL, replace it with an <img> tag
1363                         # This happened by accident in the original parser, but some people used it extensively
1364                         $img = $this->maybeMakeExternalImage( $text );
1365                         if ( $img !== false ) {
1366                                 $text = $img;
1367                         }
1368
1369                         $dtrail = '';
1370
1371                         # Set linktype for CSS - if URL==text, link is essentially free
1372                         $linktype = ($text == $url) ? 'free' : 'text';
1373
1374                         # No link text, e.g. [http://domain.tld/some.link]
1375                         if ( $text == '' ) {
1376                                 # Autonumber if allowed. See bug #5918
1377                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1378                                         $text = '[' . ++$this->mAutonumber . ']';
1379                                         $linktype = 'autonumber';
1380                                 } else {
1381                                         # Otherwise just use the URL
1382                                         $text = htmlspecialchars( $url );
1383                                         $linktype = 'free';
1384                                 }
1385                         } else {
1386                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1387                                 # Check for trail
1388                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1389                         }
1390
1391                         $text = $wgContLang->markNoConversion($text);
1392
1393                         $url = Sanitizer::cleanUrl( $url );
1394
1395                         # Process the trail (i.e. everything after this link up until start of the next link),
1396                         # replacing any non-bracketed links
1397                         $trail = $this->replaceFreeExternalLinks( $trail );
1398
1399                         # Use the encoded URL
1400                         # This means that users can paste URLs directly into the text
1401                         # Funny characters like &ouml; aren't valid in URLs anyway
1402                         # This was changed in August 2004
1403                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1404
1405                         # Register link in the output object.
1406                         # Replace unnecessary URL escape codes with the referenced character
1407                         # This prevents spammers from hiding links from the filters
1408                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1409                         $this->mOutput->addExternalLink( $pasteurized );
1410                 }
1411
1412                 wfProfileOut( $fname );
1413                 return $s;
1414         }
1415
1416         /**
1417          * Replace anything that looks like a URL with a link
1418          * @private
1419          */
1420         function replaceFreeExternalLinks( $text ) {
1421                 global $wgContLang;
1422                 $fname = 'Parser::replaceFreeExternalLinks';
1423                 wfProfileIn( $fname );
1424
1425                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1426                 $s = array_shift( $bits );
1427                 $i = 0;
1428
1429                 $sk = $this->mOptions->getSkin();
1430
1431                 while ( $i < count( $bits ) ){
1432                         $protocol = $bits[$i++];
1433                         $remainder = $bits[$i++];
1434
1435                         $m = array();
1436                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1437                                 # Found some characters after the protocol that look promising
1438                                 $url = $protocol . $m[1];
1439                                 $trail = $m[2];
1440
1441                                 # special case: handle urls as url args:
1442                                 # http://www.example.com/foo?=http://www.example.com/bar
1443                                 if(strlen($trail) == 0 &&
1444                                         isset($bits[$i]) &&
1445                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1446                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1447                                 {
1448                                         # add protocol, arg
1449                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1450                                         $i += 2;
1451                                         $trail = $m[2];
1452                                 }
1453
1454                                 # The characters '<' and '>' (which were escaped by
1455                                 # removeHTMLtags()) should not be included in
1456                                 # URLs, per RFC 2396.
1457                                 $m2 = array();
1458                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1459                                         $trail = substr($url, $m2[0][1]) . $trail;
1460                                         $url = substr($url, 0, $m2[0][1]);
1461                                 }
1462
1463                                 # Move trailing punctuation to $trail
1464                                 $sep = ',;\.:!?';
1465                                 # If there is no left bracket, then consider right brackets fair game too
1466                                 if ( strpos( $url, '(' ) === false ) {
1467                                         $sep .= ')';
1468                                 }
1469
1470                                 $numSepChars = strspn( strrev( $url ), $sep );
1471                                 if ( $numSepChars ) {
1472                                         $trail = substr( $url, -$numSepChars ) . $trail;
1473                                         $url = substr( $url, 0, -$numSepChars );
1474                                 }
1475
1476                                 $url = Sanitizer::cleanUrl( $url );
1477
1478                                 # Is this an external image?
1479                                 $text = $this->maybeMakeExternalImage( $url );
1480                                 if ( $text === false ) {
1481                                         # Not an image, make a link
1482                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1483                                         # Register it in the output object...
1484                                         # Replace unnecessary URL escape codes with their equivalent characters
1485                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1486                                         $this->mOutput->addExternalLink( $pasteurized );
1487                                 }
1488                                 $s .= $text . $trail;
1489                         } else {
1490                                 $s .= $protocol . $remainder;
1491                         }
1492                 }
1493                 wfProfileOut( $fname );
1494                 return $s;
1495         }
1496
1497         /**
1498          * Replace unusual URL escape codes with their equivalent characters
1499          * @param string
1500          * @return string
1501          * @static
1502          * @todo  This can merge genuinely required bits in the path or query string,
1503          *        breaking legit URLs. A proper fix would treat the various parts of
1504          *        the URL differently; as a workaround, just use the output for
1505          *        statistical records, not for actual linking/output.
1506          */
1507         static function replaceUnusualEscapes( $url ) {
1508                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1509                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1510         }
1511
1512         /**
1513          * Callback function used in replaceUnusualEscapes().
1514          * Replaces unusual URL escape codes with their equivalent character
1515          * @static
1516          * @private
1517          */
1518         private static function replaceUnusualEscapesCallback( $matches ) {
1519                 $char = urldecode( $matches[0] );
1520                 $ord = ord( $char );
1521                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1522                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1523                         // No, shouldn't be escaped
1524                         return $char;
1525                 } else {
1526                         // Yes, leave it escaped
1527                         return $matches[0];
1528                 }
1529         }
1530
1531         /**
1532          * make an image if it's allowed, either through the global
1533          * option or through the exception
1534          * @private
1535          */
1536         function maybeMakeExternalImage( $url ) {
1537                 $sk = $this->mOptions->getSkin();
1538                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1539                 $imagesexception = !empty($imagesfrom);
1540                 $text = false;
1541                 if ( $this->mOptions->getAllowExternalImages()
1542                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1543                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1544                                 # Image found
1545                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1546                         }
1547                 }
1548                 return $text;
1549         }
1550
1551         /**
1552          * Process [[ ]] wikilinks
1553          *
1554          * @private
1555          */
1556         function replaceInternalLinks( $s ) {
1557                 global $wgContLang;
1558                 static $fname = 'Parser::replaceInternalLinks' ;
1559
1560                 wfProfileIn( $fname );
1561
1562                 wfProfileIn( $fname.'-setup' );
1563                 static $tc = FALSE;
1564                 # the % is needed to support urlencoded titles as well
1565                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1566
1567                 $sk = $this->mOptions->getSkin();
1568
1569                 #split the entire text string on occurences of [[
1570                 $a = explode( '[[', ' ' . $s );
1571                 #get the first element (all text up to first [[), and remove the space we added
1572                 $s = array_shift( $a );
1573                 $s = substr( $s, 1 );
1574
1575                 # Match a link having the form [[namespace:link|alternate]]trail
1576                 static $e1 = FALSE;
1577                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1578                 # Match cases where there is no "]]", which might still be images
1579                 static $e1_img = FALSE;
1580                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1581                 # Match the end of a line for a word that's not followed by whitespace,
1582                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1583                 $e2 = wfMsgForContent( 'linkprefix' );
1584
1585                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1586                 if( is_null( $this->mTitle ) ) {
1587                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1588                 }
1589                 $nottalk = !$this->mTitle->isTalkPage();
1590
1591                 if ( $useLinkPrefixExtension ) {
1592                         $m = array();
1593                         if ( preg_match( $e2, $s, $m ) ) {
1594                                 $first_prefix = $m[2];
1595                         } else {
1596                                 $first_prefix = false;
1597                         }
1598                 } else {
1599                         $prefix = '';
1600                 }
1601
1602                 if($wgContLang->hasVariants()) {
1603                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1604                 } else {
1605                         $selflink = array($this->mTitle->getPrefixedText());
1606                 }
1607                 $useSubpages = $this->areSubpagesAllowed();
1608                 wfProfileOut( $fname.'-setup' );
1609
1610                 # Loop for each link
1611                 for ($k = 0; isset( $a[$k] ); $k++) {
1612                         $line = $a[$k];
1613                         if ( $useLinkPrefixExtension ) {
1614                                 wfProfileIn( $fname.'-prefixhandling' );
1615                                 if ( preg_match( $e2, $s, $m ) ) {
1616                                         $prefix = $m[2];
1617                                         $s = $m[1];
1618                                 } else {
1619                                         $prefix='';
1620                                 }
1621                                 # first link
1622                                 if($first_prefix) {
1623                                         $prefix = $first_prefix;
1624                                         $first_prefix = false;
1625                                 }
1626                                 wfProfileOut( $fname.'-prefixhandling' );
1627                         }
1628
1629                         $might_be_img = false;
1630
1631                         wfProfileIn( "$fname-e1" );
1632                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1633                                 $text = $m[2];
1634                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1635                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1636                                 # the real problem is with the $e1 regex
1637                                 # See bug 1300.
1638                                 #
1639                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1640                                 # and no image is in sight. See bug 2095.
1641                                 #
1642                                 if( $text !== '' &&
1643                                         substr( $m[3], 0, 1 ) === ']' &&
1644                                         strpos($text, '[') !== false
1645                                 )
1646                                 {
1647                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1648                                         $m[3] = substr( $m[3], 1 );
1649                                 }
1650                                 # fix up urlencoded title texts
1651                                 if( strpos( $m[1], '%' ) !== false ) {
1652                                         # Should anchors '#' also be rejected?
1653                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1654                                 }
1655                                 $trail = $m[3];
1656                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1657                                 $might_be_img = true;
1658                                 $text = $m[2];
1659                                 if ( strpos( $m[1], '%' ) !== false ) {
1660                                         $m[1] = urldecode($m[1]);
1661                                 }
1662                                 $trail = "";
1663                         } else { # Invalid form; output directly
1664                                 $s .= $prefix . '[[' . $line ;
1665                                 wfProfileOut( "$fname-e1" );
1666                                 continue;
1667                         }
1668                         wfProfileOut( "$fname-e1" );
1669                         wfProfileIn( "$fname-misc" );
1670
1671                         # Don't allow internal links to pages containing
1672                         # PROTO: where PROTO is a valid URL protocol; these
1673                         # should be external links.
1674                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1675                                 $s .= $prefix . '[[' . $line ;
1676                                 continue;
1677                         }
1678
1679                         # Make subpage if necessary
1680                         if( $useSubpages ) {
1681                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1682                         } else {
1683                                 $link = $m[1];
1684                         }
1685
1686                         $noforce = (substr($m[1], 0, 1) != ':');
1687                         if (!$noforce) {
1688                                 # Strip off leading ':'
1689                                 $link = substr($link, 1);
1690                         }
1691
1692                         wfProfileOut( "$fname-misc" );
1693                         wfProfileIn( "$fname-title" );
1694                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1695                         if( !$nt ) {
1696                                 $s .= $prefix . '[[' . $line;
1697                                 wfProfileOut( "$fname-title" );
1698                                 continue;
1699                         }
1700
1701                         $ns = $nt->getNamespace();
1702                         $iw = $nt->getInterWiki();
1703                         wfProfileOut( "$fname-title" );
1704
1705                         if ($might_be_img) { # if this is actually an invalid link
1706                                 wfProfileIn( "$fname-might_be_img" );
1707                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1708                                         $found = false;
1709                                         while (isset ($a[$k+1]) ) {
1710                                                 #look at the next 'line' to see if we can close it there
1711                                                 $spliced = array_splice( $a, $k + 1, 1 );
1712                                                 $next_line = array_shift( $spliced );
1713                                                 $m = explode( ']]', $next_line, 3 );
1714                                                 if ( count( $m ) == 3 ) {
1715                                                         # the first ]] closes the inner link, the second the image
1716                                                         $found = true;
1717                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1718                                                         $trail = $m[2];
1719                                                         break;
1720                                                 } elseif ( count( $m ) == 2 ) {
1721                                                         #if there's exactly one ]] that's fine, we'll keep looking
1722                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1723                                                 } else {
1724                                                         #if $next_line is invalid too, we need look no further
1725                                                         $text .= '[[' . $next_line;
1726                                                         break;
1727                                                 }
1728                                         }
1729                                         if ( !$found ) {
1730                                                 # we couldn't find the end of this imageLink, so output it raw
1731                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1732                                                 $text = $this->replaceInternalLinks($text);
1733                                                 $s .= "{$prefix}[[$link|$text";
1734                                                 # note: no $trail, because without an end, there *is* no trail
1735                                                 wfProfileOut( "$fname-might_be_img" );
1736                                                 continue;
1737                                         }
1738                                 } else { #it's not an image, so output it raw
1739                                         $s .= "{$prefix}[[$link|$text";
1740                                         # note: no $trail, because without an end, there *is* no trail
1741                                         wfProfileOut( "$fname-might_be_img" );
1742                                         continue;
1743                                 }
1744                                 wfProfileOut( "$fname-might_be_img" );
1745                         }
1746
1747                         $wasblank = ( '' == $text );
1748                         if( $wasblank ) $text = $link;
1749
1750                         # Link not escaped by : , create the various objects
1751                         if( $noforce ) {
1752
1753                                 # Interwikis
1754                                 wfProfileIn( "$fname-interwiki" );
1755                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1756                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1757                                         $s = rtrim($s . $prefix);
1758                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1759                                         wfProfileOut( "$fname-interwiki" );
1760                                         continue;
1761                                 }
1762                                 wfProfileOut( "$fname-interwiki" );
1763
1764                                 if ( $ns == NS_IMAGE ) {
1765                                         wfProfileIn( "$fname-image" );
1766                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1767                                                 # recursively parse links inside the image caption
1768                                                 # actually, this will parse them in any other parameters, too,
1769                                                 # but it might be hard to fix that, and it doesn't matter ATM
1770                                                 $text = $this->replaceExternalLinks($text);
1771                                                 $text = $this->replaceInternalLinks($text);
1772
1773                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1774                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1775                                                 $this->mOutput->addImage( $nt->getDBkey() );
1776
1777                                                 wfProfileOut( "$fname-image" );
1778                                                 continue;
1779                                         } else {
1780                                                 # We still need to record the image's presence on the page
1781                                                 $this->mOutput->addImage( $nt->getDBkey() );
1782                                         }
1783                                         wfProfileOut( "$fname-image" );
1784
1785                                 }
1786
1787                                 if ( $ns == NS_CATEGORY ) {
1788                                         wfProfileIn( "$fname-category" );
1789                                         $s = rtrim($s . "\n"); # bug 87
1790
1791                                         if ( $wasblank ) {
1792                                                 $sortkey = $this->getDefaultSort();
1793                                         } else {
1794                                                 $sortkey = $text;
1795                                         }
1796                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1797                                         $sortkey = str_replace( "\n", '', $sortkey );
1798                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1799                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1800
1801                                         /**
1802                                          * Strip the whitespace Category links produce, see bug 87
1803                                          * @todo We might want to use trim($tmp, "\n") here.
1804                                          */
1805                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1806
1807                                         wfProfileOut( "$fname-category" );
1808                                         continue;
1809                                 }
1810                         }
1811
1812                         # Self-link checking
1813                         if( $nt->getFragment() === '' ) {
1814                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1815                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1816                                         continue;
1817                                 }
1818                         }
1819
1820                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1821                         if( $ns == NS_MEDIA ) {
1822                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1823                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1824                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1825                                 $this->mOutput->addImage( $nt->getDBkey() );
1826                                 continue;
1827                         } elseif( $ns == NS_SPECIAL ) {
1828                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1829                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1830                                 } else {
1831                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1832                                 }
1833                                 continue;
1834                         } elseif( $ns == NS_IMAGE ) {
1835                                 $img = wfFindFile( $nt );
1836                                 if( $img ) {
1837                                         // Force a blue link if the file exists; may be a remote
1838                                         // upload on the shared repository, and we want to see its
1839                                         // auto-generated page.
1840                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1841                                         $this->mOutput->addLink( $nt );
1842                                         continue;
1843                                 }
1844                         }
1845                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1846                 }
1847                 wfProfileOut( $fname );
1848                 return $s;
1849         }
1850
1851         /**
1852          * Make a link placeholder. The text returned can be later resolved to a real link with
1853          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1854          * parsing of interwiki links, and secondly to allow all existence checks and
1855          * article length checks (for stub links) to be bundled into a single query.
1856          *
1857          */
1858         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1859                 wfProfileIn( __METHOD__ );
1860                 if ( ! is_object($nt) ) {
1861                         # Fail gracefully
1862                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1863                 } else {
1864                         # Separate the link trail from the rest of the link
1865                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1866
1867                         if ( $nt->isExternal() ) {
1868                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1869                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1870                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1871                         } else {
1872                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1873                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1874                                 $this->mLinkHolders['queries'][] = $query;
1875                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1876                                 $this->mLinkHolders['titles'][] = $nt;
1877
1878                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1879                         }
1880                 }
1881                 wfProfileOut( __METHOD__ );
1882                 return $retVal;
1883         }
1884
1885         /**
1886          * Render a forced-blue link inline; protect against double expansion of
1887          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1888          * Since this little disaster has to split off the trail text to avoid
1889          * breaking URLs in the following text without breaking trails on the
1890          * wiki links, it's been made into a horrible function.
1891          *
1892          * @param Title $nt
1893          * @param string $text
1894          * @param string $query
1895          * @param string $trail
1896          * @param string $prefix
1897          * @return string HTML-wikitext mix oh yuck
1898          */
1899         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1900                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1901                 $sk = $this->mOptions->getSkin();
1902                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1903                 return $this->armorLinks( $link ) . $trail;
1904         }
1905
1906         /**
1907          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1908          * going to go through further parsing steps before inline URL expansion.
1909          *
1910          * In particular this is important when using action=render, which causes
1911          * full URLs to be included.
1912          *
1913          * Oh man I hate our multi-layer parser!
1914          *
1915          * @param string more-or-less HTML
1916          * @return string less-or-more HTML with NOPARSE bits
1917          */
1918         function armorLinks( $text ) {
1919                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1920                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1921         }
1922
1923         /**
1924          * Return true if subpage links should be expanded on this page.
1925          * @return bool
1926          */
1927         function areSubpagesAllowed() {
1928                 # Some namespaces don't allow subpages
1929                 global $wgNamespacesWithSubpages;
1930                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1931         }
1932
1933         /**
1934          * Handle link to subpage if necessary
1935          * @param string $target the source of the link
1936          * @param string &$text the link text, modified as necessary
1937          * @return string the full name of the link
1938          * @private
1939          */
1940         function maybeDoSubpageLink($target, &$text) {
1941                 # Valid link forms:
1942                 # Foobar -- normal
1943                 # :Foobar -- override special treatment of prefix (images, language links)
1944                 # /Foobar -- convert to CurrentPage/Foobar
1945                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1946                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1947                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1948
1949                 $fname = 'Parser::maybeDoSubpageLink';
1950                 wfProfileIn( $fname );
1951                 $ret = $target; # default return value is no change
1952
1953                 # Some namespaces don't allow subpages,
1954                 # so only perform processing if subpages are allowed
1955                 if( $this->areSubpagesAllowed() ) {
1956                         $hash = strpos( $target, '#' );
1957                         if( $hash !== false ) {
1958                                 $suffix = substr( $target, $hash );
1959                                 $target = substr( $target, 0, $hash );
1960                         } else {
1961                                 $suffix = '';
1962                         }
1963                         # bug 7425
1964                         $target = trim( $target );
1965                         # Look at the first character
1966                         if( $target != '' && $target{0} == '/' ) {
1967                                 # / at end means we don't want the slash to be shown
1968                                 $m = array();
1969                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1970                                 if( $trailingSlashes ) {
1971                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1972                                 } else {
1973                                         $noslash = substr( $target, 1 );
1974                                 }
1975
1976                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1977                                 if( '' === $text ) {
1978                                         $text = $target . $suffix;
1979                                 } # this might be changed for ugliness reasons
1980                         } else {
1981                                 # check for .. subpage backlinks
1982                                 $dotdotcount = 0;
1983                                 $nodotdot = $target;
1984                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1985                                         ++$dotdotcount;
1986                                         $nodotdot = substr( $nodotdot, 3 );
1987                                 }
1988                                 if($dotdotcount > 0) {
1989                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1990                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1991                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1992                                                 # / at the end means don't show full path
1993                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1994                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1995                                                         if( '' === $text ) {
1996                                                                 $text = $nodotdot . $suffix;
1997                                                         }
1998                                                 }
1999                                                 $nodotdot = trim( $nodotdot );
2000                                                 if( $nodotdot != '' ) {
2001                                                         $ret .= '/' . $nodotdot;
2002                                                 }
2003                                                 $ret .= $suffix;
2004                                         }
2005                                 }
2006                         }
2007                 }
2008
2009                 wfProfileOut( $fname );
2010                 return $ret;
2011         }
2012
2013         /**#@+
2014          * Used by doBlockLevels()
2015          * @private
2016          */
2017         /* private */ function closeParagraph() {
2018                 $result = '';
2019                 if ( '' != $this->mLastSection ) {
2020                         $result = '</' . $this->mLastSection  . ">\n";
2021                 }
2022                 $this->mInPre = false;
2023                 $this->mLastSection = '';
2024                 return $result;
2025         }
2026         # getCommon() returns the length of the longest common substring
2027         # of both arguments, starting at the beginning of both.
2028         #
2029         /* private */ function getCommon( $st1, $st2 ) {
2030                 $fl = strlen( $st1 );
2031                 $shorter = strlen( $st2 );
2032                 if ( $fl < $shorter ) { $shorter = $fl; }
2033
2034                 for ( $i = 0; $i < $shorter; ++$i ) {
2035                         if ( $st1{$i} != $st2{$i} ) { break; }
2036                 }
2037                 return $i;
2038         }
2039         # These next three functions open, continue, and close the list
2040         # element appropriate to the prefix character passed into them.
2041         #
2042         /* private */ function openList( $char ) {
2043                 $result = $this->closeParagraph();
2044
2045                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2046                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2047                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2048                 else if ( ';' == $char ) {
2049                         $result .= '<dl><dt>';
2050                         $this->mDTopen = true;
2051                 }
2052                 else { $result = '<!-- ERR 1 -->'; }
2053
2054                 return $result;
2055         }
2056
2057         /* private */ function nextItem( $char ) {
2058                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2059                 else if ( ':' == $char || ';' == $char ) {
2060                         $close = '</dd>';
2061                         if ( $this->mDTopen ) { $close = '</dt>'; }
2062                         if ( ';' == $char ) {
2063                                 $this->mDTopen = true;
2064                                 return $close . '<dt>';
2065                         } else {
2066                                 $this->mDTopen = false;
2067                                 return $close . '<dd>';
2068                         }
2069                 }
2070                 return '<!-- ERR 2 -->';
2071         }
2072
2073         /* private */ function closeList( $char ) {
2074                 if ( '*' == $char ) { $text = '</li></ul>'; }
2075                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2076                 else if ( ':' == $char ) {
2077                         if ( $this->mDTopen ) {
2078                                 $this->mDTopen = false;
2079                                 $text = '</dt></dl>';
2080                         } else {
2081                                 $text = '</dd></dl>';
2082                         }
2083                 }
2084                 else {  return '<!-- ERR 3 -->'; }
2085                 return $text."\n";
2086         }
2087         /**#@-*/
2088
2089         /**
2090          * Make lists from lines starting with ':', '*', '#', etc.
2091          *
2092          * @private
2093          * @return string the lists rendered as HTML
2094          */
2095         function doBlockLevels( $text, $linestart ) {
2096                 $fname = 'Parser::doBlockLevels';
2097                 wfProfileIn( $fname );
2098
2099                 # Parsing through the text line by line.  The main thing
2100                 # happening here is handling of block-level elements p, pre,
2101                 # and making lists from lines starting with * # : etc.
2102                 #
2103                 $textLines = explode( "\n", $text );
2104
2105                 $lastPrefix = $output = '';
2106                 $this->mDTopen = $inBlockElem = false;
2107                 $prefixLength = 0;
2108                 $paragraphStack = false;
2109
2110                 if ( !$linestart ) {
2111                         $output .= array_shift( $textLines );
2112                 }
2113                 foreach ( $textLines as $oLine ) {
2114                         $lastPrefixLength = strlen( $lastPrefix );
2115                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2116                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2117                         if ( !$this->mInPre ) {
2118                                 # Multiple prefixes may abut each other for nested lists.
2119                                 $prefixLength = strspn( $oLine, '*#:;' );
2120                                 $pref = substr( $oLine, 0, $prefixLength );
2121
2122                                 # eh?
2123                                 $pref2 = str_replace( ';', ':', $pref );
2124                                 $t = substr( $oLine, $prefixLength );
2125                                 $this->mInPre = !empty($preOpenMatch);
2126                         } else {
2127                                 # Don't interpret any other prefixes in preformatted text
2128                                 $prefixLength = 0;
2129                                 $pref = $pref2 = '';
2130                                 $t = $oLine;
2131                         }
2132
2133                         # List generation
2134                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2135                                 # Same as the last item, so no need to deal with nesting or opening stuff
2136                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2137                                 $paragraphStack = false;
2138
2139                                 if ( substr( $pref, -1 ) == ';') {
2140                                         # The one nasty exception: definition lists work like this:
2141                                         # ; title : definition text
2142                                         # So we check for : in the remainder text to split up the
2143                                         # title and definition, without b0rking links.
2144                                         $term = $t2 = '';
2145                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2146                                                 $t = $t2;
2147                                                 $output .= $term . $this->nextItem( ':' );
2148                                         }
2149                                 }
2150                         } elseif( $prefixLength || $lastPrefixLength ) {
2151                                 # Either open or close a level...
2152                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2153                                 $paragraphStack = false;
2154
2155                                 while( $commonPrefixLength < $lastPrefixLength ) {
2156                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2157                                         --$lastPrefixLength;
2158                                 }
2159                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2160                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2161                                 }
2162                                 while ( $prefixLength > $commonPrefixLength ) {
2163                                         $char = substr( $pref, $commonPrefixLength, 1 );
2164                                         $output .= $this->openList( $char );
2165
2166                                         if ( ';' == $char ) {
2167                                                 # FIXME: This is dupe of code above
2168                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2169                                                         $t = $t2;
2170                                                         $output .= $term . $this->nextItem( ':' );
2171                                                 }
2172                                         }
2173                                         ++$commonPrefixLength;
2174                                 }
2175                                 $lastPrefix = $pref2;
2176                         }
2177                         if( 0 == $prefixLength ) {
2178                                 wfProfileIn( "$fname-paragraph" );
2179                                 # No prefix (not in list)--go to paragraph mode
2180                                 // XXX: use a stack for nestable elements like span, table and div
2181                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2182                                 $closematch = preg_match(
2183                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2184                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2185                                 if ( $openmatch or $closematch ) {
2186                                         $paragraphStack = false;
2187                                         # TODO bug 5718: paragraph closed
2188                                         $output .= $this->closeParagraph();
2189                                         if ( $preOpenMatch and !$preCloseMatch ) {
2190                                                 $this->mInPre = true;
2191                                         }
2192                                         if ( $closematch ) {
2193                                                 $inBlockElem = false;
2194                                         } else {
2195                                                 $inBlockElem = true;
2196                                         }
2197                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2198                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2199                                                 // pre
2200                                                 if ($this->mLastSection != 'pre') {
2201                                                         $paragraphStack = false;
2202                                                         $output .= $this->closeParagraph().'<pre>';
2203                                                         $this->mLastSection = 'pre';
2204                                                 }
2205                                                 $t = substr( $t, 1 );
2206                                         } else {
2207                                                 // paragraph
2208                                                 if ( '' == trim($t) ) {
2209                                                         if ( $paragraphStack ) {
2210                                                                 $output .= $paragraphStack.'<br />';
2211                                                                 $paragraphStack = false;
2212                                                                 $this->mLastSection = 'p';
2213                                                         } else {
2214                                                                 if ($this->mLastSection != 'p' ) {
2215                                                                         $output .= $this->closeParagraph();
2216                                                                         $this->mLastSection = '';
2217                                                                         $paragraphStack = '<p>';
2218                                                                 } else {
2219                                                                         $paragraphStack = '</p><p>';
2220                                                                 }
2221                                                         }
2222                                                 } else {
2223                                                         if ( $paragraphStack ) {
2224                                                                 $output .= $paragraphStack;
2225                                                                 $paragraphStack = false;
2226                                                                 $this->mLastSection = 'p';
2227                                                         } else if ($this->mLastSection != 'p') {
2228                                                                 $output .= $this->closeParagraph().'<p>';
2229                                                                 $this->mLastSection = 'p';
2230                                                         }
2231                                                 }
2232                                         }
2233                                 }
2234                                 wfProfileOut( "$fname-paragraph" );
2235                         }
2236                         // somewhere above we forget to get out of pre block (bug 785)
2237                         if($preCloseMatch && $this->mInPre) {
2238                                 $this->mInPre = false;
2239                         }
2240                         if ($paragraphStack === false) {
2241                                 $output .= $t."\n";
2242                         }
2243                 }
2244                 while ( $prefixLength ) {
2245                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2246                         --$prefixLength;
2247                 }
2248                 if ( '' != $this->mLastSection ) {
2249                         $output .= '</' . $this->mLastSection . '>';
2250                         $this->mLastSection = '';
2251                 }
2252
2253                 wfProfileOut( $fname );
2254                 return $output;
2255         }
2256
2257         /**
2258          * Split up a string on ':', ignoring any occurences inside tags
2259          * to prevent illegal overlapping.
2260          * @param string $str the string to split
2261          * @param string &$before set to everything before the ':'
2262          * @param string &$after set to everything after the ':'
2263          * return string the position of the ':', or false if none found
2264          */
2265         function findColonNoLinks($str, &$before, &$after) {
2266                 $fname = 'Parser::findColonNoLinks';
2267                 wfProfileIn( $fname );
2268
2269                 $pos = strpos( $str, ':' );
2270                 if( $pos === false ) {
2271                         // Nothing to find!
2272                         wfProfileOut( $fname );
2273                         return false;
2274                 }
2275
2276                 $lt = strpos( $str, '<' );
2277                 if( $lt === false || $lt > $pos ) {
2278                         // Easy; no tag nesting to worry about
2279                         $before = substr( $str, 0, $pos );
2280                         $after = substr( $str, $pos+1 );
2281                         wfProfileOut( $fname );
2282                         return $pos;
2283                 }
2284
2285                 // Ugly state machine to walk through avoiding tags.
2286                 $state = MW_COLON_STATE_TEXT;
2287                 $stack = 0;
2288                 $len = strlen( $str );
2289                 for( $i = 0; $i < $len; $i++ ) {
2290                         $c = $str{$i};
2291
2292                         switch( $state ) {
2293                         // (Using the number is a performance hack for common cases)
2294                         case 0: // MW_COLON_STATE_TEXT:
2295                                 switch( $c ) {
2296                                 case "<":
2297                                         // Could be either a <start> tag or an </end> tag
2298                                         $state = MW_COLON_STATE_TAGSTART;
2299                                         break;
2300                                 case ":":
2301                                         if( $stack == 0 ) {
2302                                                 // We found it!
2303                                                 $before = substr( $str, 0, $i );
2304                                                 $after = substr( $str, $i + 1 );
2305                                                 wfProfileOut( $fname );
2306                                                 return $i;
2307                                         }
2308                                         // Embedded in a tag; don't break it.
2309                                         break;
2310                                 default:
2311                                         // Skip ahead looking for something interesting
2312                                         $colon = strpos( $str, ':', $i );
2313                                         if( $colon === false ) {
2314                                                 // Nothing else interesting
2315                                                 wfProfileOut( $fname );
2316                                                 return false;
2317                                         }
2318                                         $lt = strpos( $str, '<', $i );
2319                                         if( $stack === 0 ) {
2320                                                 if( $lt === false || $colon < $lt ) {
2321                                                         // We found it!
2322                                                         $before = substr( $str, 0, $colon );
2323                                                         $after = substr( $str, $colon + 1 );
2324                                                         wfProfileOut( $fname );
2325                                                         return $i;
2326                                                 }
2327                                         }
2328                                         if( $lt === false ) {
2329                                                 // Nothing else interesting to find; abort!
2330                                                 // We're nested, but there's no close tags left. Abort!
2331                                                 break 2;
2332                                         }
2333                                         // Skip ahead to next tag start
2334                                         $i = $lt;
2335                                         $state = MW_COLON_STATE_TAGSTART;
2336                                 }
2337                                 break;
2338                         case 1: // MW_COLON_STATE_TAG:
2339                                 // In a <tag>
2340                                 switch( $c ) {
2341                                 case ">":
2342                                         $stack++;
2343                                         $state = MW_COLON_STATE_TEXT;
2344                                         break;
2345                                 case "/":
2346                                         // Slash may be followed by >?
2347                                         $state = MW_COLON_STATE_TAGSLASH;
2348                                         break;
2349                                 default:
2350                                         // ignore
2351                                 }
2352                                 break;
2353                         case 2: // MW_COLON_STATE_TAGSTART:
2354                                 switch( $c ) {
2355                                 case "/":
2356                                         $state = MW_COLON_STATE_CLOSETAG;
2357                                         break;
2358                                 case "!":
2359                                         $state = MW_COLON_STATE_COMMENT;
2360                                         break;
2361                                 case ">":
2362                                         // Illegal early close? This shouldn't happen D:
2363                                         $state = MW_COLON_STATE_TEXT;
2364                                         break;
2365                                 default:
2366                                         $state = MW_COLON_STATE_TAG;
2367                                 }
2368                                 break;
2369                         case 3: // MW_COLON_STATE_CLOSETAG:
2370                                 // In a </tag>
2371                                 if( $c == ">" ) {
2372                                         $stack--;
2373                                         if( $stack < 0 ) {
2374                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2375                                                 wfProfileOut( $fname );
2376                                                 return false;
2377                                         }
2378                                         $state = MW_COLON_STATE_TEXT;
2379                                 }
2380                                 break;
2381                         case MW_COLON_STATE_TAGSLASH:
2382                                 if( $c == ">" ) {
2383                                         // Yes, a self-closed tag <blah/>
2384                                         $state = MW_COLON_STATE_TEXT;
2385                                 } else {
2386                                         // Probably we're jumping the gun, and this is an attribute
2387                                         $state = MW_COLON_STATE_TAG;
2388                                 }
2389                                 break;
2390                         case 5: // MW_COLON_STATE_COMMENT:
2391                                 if( $c == "-" ) {
2392                                         $state = MW_COLON_STATE_COMMENTDASH;
2393                                 }
2394                                 break;
2395                         case MW_COLON_STATE_COMMENTDASH:
2396                                 if( $c == "-" ) {
2397                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2398                                 } else {
2399                                         $state = MW_COLON_STATE_COMMENT;
2400                                 }
2401                                 break;
2402                         case MW_COLON_STATE_COMMENTDASHDASH:
2403                                 if( $c == ">" ) {
2404                                         $state = MW_COLON_STATE_TEXT;
2405                                 } else {
2406                                         $state = MW_COLON_STATE_COMMENT;
2407                                 }
2408                                 break;
2409                         default:
2410                                 throw new MWException( "State machine error in $fname" );
2411                         }
2412                 }
2413                 if( $stack > 0 ) {
2414                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2415                         return false;
2416                 }
2417                 wfProfileOut( $fname );
2418                 return false;
2419         }
2420
2421         /**
2422          * Return value of a magic variable (like PAGENAME)
2423          *
2424          * @private
2425          */
2426         function getVariableValue( $index ) {
2427                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2428
2429                 /**
2430                  * Some of these require message or data lookups and can be
2431                  * expensive to check many times.
2432                  */
2433                 static $varCache = array();
2434                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2435                         if ( isset( $varCache[$index] ) ) {
2436                                 return $varCache[$index];
2437                         }
2438                 }
2439
2440                 $ts = time();
2441                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2442
2443                 # Use the time zone
2444                 global $wgLocaltimezone;
2445                 if ( isset( $wgLocaltimezone ) ) {
2446                         $oldtz = getenv( 'TZ' );
2447                         putenv( 'TZ='.$wgLocaltimezone );
2448                 }
2449
2450                 wfSuppressWarnings(); // E_STRICT system time bitching
2451                 $localTimestamp = date( 'YmdHis', $ts );
2452                 $localMonth = date( 'm', $ts );
2453                 $localMonthName = date( 'n', $ts );
2454                 $localDay = date( 'j', $ts );
2455                 $localDay2 = date( 'd', $ts );
2456                 $localDayOfWeek = date( 'w', $ts );
2457                 $localWeek = date( 'W', $ts );
2458                 $localYear = date( 'Y', $ts );
2459                 $localHour = date( 'H', $ts );
2460                 if ( isset( $wgLocaltimezone ) ) {
2461                         putenv( 'TZ='.$oldtz );
2462                 }
2463                 wfRestoreWarnings();
2464
2465                 switch ( $index ) {
2466                         case 'currentmonth':
2467                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2468                         case 'currentmonthname':
2469                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2470                         case 'currentmonthnamegen':
2471                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2472                         case 'currentmonthabbrev':
2473                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2474                         case 'currentday':
2475                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2476                         case 'currentday2':
2477                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2478                         case 'localmonth':
2479                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2480                         case 'localmonthname':
2481                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2482                         case 'localmonthnamegen':
2483                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2484                         case 'localmonthabbrev':
2485                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2486                         case 'localday':
2487                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2488                         case 'localday2':
2489                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2490                         case 'pagename':
2491                                 return wfEscapeWikiText( $this->mTitle->getText() );
2492                         case 'pagenamee':
2493                                 return $this->mTitle->getPartialURL();
2494                         case 'fullpagename':
2495                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2496                         case 'fullpagenamee':
2497                                 return $this->mTitle->getPrefixedURL();
2498                         case 'subpagename':
2499                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2500                         case 'subpagenamee':
2501                                 return $this->mTitle->getSubpageUrlForm();
2502                         case 'basepagename':
2503                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2504                         case 'basepagenamee':
2505                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2506                         case 'talkpagename':
2507                                 if( $this->mTitle->canTalk() ) {
2508                                         $talkPage = $this->mTitle->getTalkPage();
2509                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2510                                 } else {
2511                                         return '';
2512                                 }
2513                         case 'talkpagenamee':
2514                                 if( $this->mTitle->canTalk() ) {
2515                                         $talkPage = $this->mTitle->getTalkPage();
2516                                         return $talkPage->getPrefixedUrl();
2517                                 } else {
2518                                         return '';
2519                                 }
2520                         case 'subjectpagename':
2521                                 $subjPage = $this->mTitle->getSubjectPage();
2522                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2523                         case 'subjectpagenamee':
2524                                 $subjPage = $this->mTitle->getSubjectPage();
2525                                 return $subjPage->getPrefixedUrl();
2526                         case 'revisionid':
2527                                 return $this->mRevisionId;
2528                         case 'revisionday':
2529                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2530                         case 'revisionday2':
2531                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2532                         case 'revisionmonth':
2533                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2534                         case 'revisionyear':
2535                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2536                         case 'revisiontimestamp':
2537                                 return $this->getRevisionTimestamp();
2538                         case 'namespace':
2539                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2540                         case 'namespacee':
2541                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2542                         case 'talkspace':
2543                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2544                         case 'talkspacee':
2545                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2546                         case 'subjectspace':
2547                                 return $this->mTitle->getSubjectNsText();
2548                         case 'subjectspacee':
2549                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2550                         case 'currentdayname':
2551                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2552                         case 'currentyear':
2553                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2554                         case 'currenttime':
2555                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2556                         case 'currenthour':
2557                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2558                         case 'currentweek':
2559                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2560                                 // int to remove the padding
2561                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2562                         case 'currentdow':
2563                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2564                         case 'localdayname':
2565                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2566                         case 'localyear':
2567                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2568                         case 'localtime':
2569                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2570                         case 'localhour':
2571                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2572                         case 'localweek':
2573                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2574                                 // int to remove the padding
2575                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2576                         case 'localdow':
2577                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2578                         case 'numberofarticles':
2579                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2580                         case 'numberoffiles':
2581                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2582                         case 'numberofusers':
2583                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2584                         case 'numberofpages':
2585                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2586                         case 'numberofadmins':
2587                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2588                         case 'numberofedits':
2589                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2590                         case 'currenttimestamp':
2591                                 return $varCache[$index] = wfTimestampNow();
2592                         case 'localtimestamp':
2593                                 return $varCache[$index] = $localTimestamp;
2594                         case 'currentversion':
2595                                 return $varCache[$index] = SpecialVersion::getVersion();
2596                         case 'sitename':
2597                                 return $wgSitename;
2598                         case 'server':
2599                                 return $wgServer;
2600                         case 'servername':
2601                                 return $wgServerName;
2602                         case 'scriptpath':
2603                                 return $wgScriptPath;
2604                         case 'directionmark':
2605                                 return $wgContLang->getDirMark();
2606                         case 'contentlanguage':
2607                                 global $wgContLanguageCode;
2608                                 return $wgContLanguageCode;
2609                         default:
2610                                 $ret = null;
2611                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2612                                         return $ret;
2613                                 else
2614                                         return null;
2615                 }
2616         }
2617
2618         /**
2619          * initialise the magic variables (like CURRENTMONTHNAME)
2620          *
2621          * @private
2622          */
2623         function initialiseVariables() {
2624                 $fname = 'Parser::initialiseVariables';
2625                 wfProfileIn( $fname );
2626                 $variableIDs = MagicWord::getVariableIDs();
2627
2628                 $this->mVariables = array();
2629                 foreach ( $variableIDs as $id ) {
2630                         $mw =& MagicWord::get( $id );
2631                         $mw->addToArray( $this->mVariables, $id );
2632                 }
2633                 wfProfileOut( $fname );
2634         }
2635
2636         /**
2637          * parse any parentheses in format ((title|part|part))
2638          * and call callbacks to get a replacement text for any found piece
2639          *
2640          * @param string $text The text to parse
2641          * @param array $callbacks rules in form:
2642          *     '{' => array(                            # opening parentheses
2643          *                                      'end' => '}',   # closing parentheses
2644          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2645          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2646          *                                                                )
2647          *                                      )
2648          *                                      'min' => 2,     # Minimum parenthesis count in cb
2649          *                                      'max' => 3,     # Maximum parenthesis count in cb
2650          * @private
2651          */
2652         function replace_callback ($text, $callbacks) {
2653                 wfProfileIn( __METHOD__ );
2654                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2655                 $lastOpeningBrace = -1;                 # last not closed parentheses
2656
2657                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2658
2659                 $i = 0;
2660                 while ( $i < strlen( $text ) ) {
2661                         # Find next opening brace, closing brace or pipe
2662                         if ( $lastOpeningBrace == -1 ) {
2663                                 $currentClosing = '';
2664                                 $search = $validOpeningBraces;
2665                         } else {
2666                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2667                                 $search = $validOpeningBraces . '|' . $currentClosing;
2668                         }
2669                         $rule = null;
2670                         $i += strcspn( $text, $search, $i );
2671                         if ( $i < strlen( $text ) ) {
2672                                 if ( $text[$i] == '|' ) {
2673                                         $found = 'pipe';
2674                                 } elseif ( $text[$i] == $currentClosing ) {
2675                                         $found = 'close';
2676                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2677                                         $found = 'open';
2678                                         $rule = $callbacks[$text[$i]];
2679                                 } else {
2680                                         # Some versions of PHP have a strcspn which stops on null characters
2681                                         # Ignore and continue
2682                                         ++$i;
2683                                         continue;
2684                                 }
2685                         } else {
2686                                 # All done
2687                                 break;
2688                         }
2689
2690                         if ( $found == 'open' ) {
2691                                 # found opening brace, let's add it to parentheses stack
2692                                 $piece = array('brace' => $text[$i],
2693                                                            'braceEnd' => $rule['end'],
2694                                                            'title' => '',
2695                                                            'parts' => null);
2696
2697                                 # count opening brace characters
2698                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2699                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2700                                 $i += $piece['count'];
2701
2702                                 # we need to add to stack only if opening brace count is enough for one of the rules
2703                                 if ( $piece['count'] >= $rule['min'] ) {
2704                                         $lastOpeningBrace ++;
2705                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2706                                 }
2707                         } elseif ( $found == 'close' ) {
2708                                 # lets check if it is enough characters for closing brace
2709                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2710                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2711
2712                                 # check for maximum matching characters (if there are 5 closing
2713                                 # characters, we will probably need only 3 - depending on the rules)
2714                                 $matchingCount = 0;
2715                                 $matchingCallback = null;
2716                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2717                                 if ( $count > $cbType['max'] ) {
2718                                         # The specified maximum exists in the callback array, unless the caller
2719                                         # has made an error
2720                                         $matchingCount = $cbType['max'];
2721                                 } else {
2722                                         # Count is less than the maximum
2723                                         # Skip any gaps in the callback array to find the true largest match
2724                                         # Need to use array_key_exists not isset because the callback can be null
2725                                         $matchingCount = $count;
2726                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2727                                                 --$matchingCount;
2728                                         }
2729                                 }
2730
2731                                 if ($matchingCount <= 0) {
2732                                         $i += $count;
2733                                         continue;
2734                                 }
2735                                 $matchingCallback = $cbType['cb'][$matchingCount];
2736
2737                                 # let's set a title or last part (if '|' was found)
2738                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2739                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2740                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2741                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2742                                 } else {
2743                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2744                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2745                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2746                                 }
2747
2748                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2749                                 $pieceEnd = $i + $matchingCount;
2750
2751                                 if( is_callable( $matchingCallback ) ) {
2752                                         $cbArgs = array (
2753                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2754                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2755                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2756                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2757                                                                          );
2758                                         # finally we can call a user callback and replace piece of text
2759                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2760                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2761                                         $i = $pieceStart + strlen($replaceWith);
2762                                 } else {
2763                                         # null value for callback means that parentheses should be parsed, but not replaced
2764                                         $i += $matchingCount;
2765                                 }
2766
2767                                 # reset last opening parentheses, but keep it in case there are unused characters
2768                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2769                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2770                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2771                                                            'title' => '',
2772                                                            'parts' => null,
2773                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2774                                 $openingBraceStack[$lastOpeningBrace--] = null;
2775
2776                                 if ($matchingCount < $piece['count']) {
2777                                         $piece['count'] -= $matchingCount;
2778                                         $piece['startAt'] -= $matchingCount;
2779                                         $piece['partStart'] = $piece['startAt'];
2780                                         # do we still qualify for any callback with remaining count?
2781                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2782                                         while ( $piece['count'] ) {
2783                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2784                                                         $lastOpeningBrace++;
2785                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2786                                                         break;
2787                                                 }
2788                                                 --$piece['count'];
2789                                         }
2790                                 }
2791                         } elseif ( $found == 'pipe' ) {
2792                                 # lets set a title if it is a first separator, or next part otherwise
2793                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2794                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2795                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2796                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2797                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2798                                 } else {
2799                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2800                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2801                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2802                                 }
2803                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2804                         }
2805                 }
2806
2807                 wfProfileOut( __METHOD__ );
2808                 return $text;
2809         }
2810
2811         /**
2812          * Replace magic variables, templates, and template arguments
2813          * with the appropriate text. Templates are substituted recursively,
2814          * taking care to avoid infinite loops.
2815          *
2816          * Note that the substitution depends on value of $mOutputType:
2817          *  OT_WIKI: only {{subst:}} templates
2818          *  OT_MSG: only magic variables
2819          *  OT_HTML: all templates and magic variables
2820          *
2821          * @param string $tex The text to transform
2822          * @param array $args Key-value pairs representing template parameters to substitute
2823          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2824          * @private
2825          */
2826         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2827                 # Prevent too big inclusions
2828                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2829                         return $text;
2830                 }
2831
2832                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2833                 wfProfileIn( $fname );
2834
2835                 # This function is called recursively. To keep track of arguments we need a stack:
2836                 array_push( $this->mArgStack, $args );
2837
2838                 $braceCallbacks = array();
2839                 if ( !$argsOnly ) {
2840                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2841                 }
2842                 if ( $this->mOutputType != OT_MSG ) {
2843                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2844                 }
2845                 if ( $braceCallbacks ) {
2846                         $callbacks = array(
2847                                 '{' => array(
2848                                         'end' => '}',
2849                                         'cb' => $braceCallbacks,
2850                                         'min' => $argsOnly ? 3 : 2,
2851                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2852                                 ),
2853                                 '[' => array(
2854                                         'end' => ']',
2855                                         'cb' => array(2=>null),
2856                                         'min' => 2,
2857                                         'max' => 2,
2858                                 )
2859                         );
2860                         $text = $this->replace_callback ($text, $callbacks);
2861
2862                         array_pop( $this->mArgStack );
2863                 }
2864                 wfProfileOut( $fname );
2865                 return $text;
2866         }
2867
2868         /**
2869          * Replace magic variables
2870          * @private
2871          */
2872         function variableSubstitution( $matches ) {
2873                 global $wgContLang;
2874                 $fname = 'Parser::variableSubstitution';
2875                 $varname = $wgContLang->lc($matches[1]);
2876                 wfProfileIn( $fname );
2877                 $skip = false;
2878                 if ( $this->mOutputType == OT_WIKI ) {
2879                         # Do only magic variables prefixed by SUBST
2880                         $mwSubst =& MagicWord::get( 'subst' );
2881                         if (!$mwSubst->matchStartAndRemove( $varname ))
2882                                 $skip = true;
2883                         # Note that if we don't substitute the variable below,
2884                         # we don't remove the {{subst:}} magic word, in case
2885                         # it is a template rather than a magic variable.
2886                 }
2887                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2888                         $id = $this->mVariables[$varname];
2889                         # Now check if we did really match, case sensitive or not
2890                         $mw =& MagicWord::get( $id );
2891                         if ($mw->match($matches[1])) {
2892                                 $text = $this->getVariableValue( $id );
2893                                 $this->mOutput->mContainsOldMagic = true;
2894                         } else {
2895                                 $text = $matches[0];
2896                         }
2897                 } else {
2898                         $text = $matches[0];
2899                 }
2900                 wfProfileOut( $fname );
2901                 return $text;
2902         }
2903
2904
2905         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2906         static function createAssocArgs( $args ) {
2907                 $assocArgs = array();
2908                 $index = 1;
2909                 foreach( $args as $arg ) {
2910                         $eqpos = strpos( $arg, '=' );
2911                         if ( $eqpos === false ) {
2912                                 $assocArgs[$index++] = $arg;
2913                         } else {
2914                                 $name = trim( substr( $arg, 0, $eqpos ) );
2915                                 $value = trim( substr( $arg, $eqpos+1 ) );
2916                                 if ( $value === false ) {
2917                                         $value = '';
2918                                 }
2919                                 if ( $name !== false ) {
2920                                         $assocArgs[$name] = $value;
2921                                 }
2922                         }
2923                 }
2924
2925                 return $assocArgs;
2926         }
2927
2928         /**
2929          * Return the text of a template, after recursively
2930          * replacing any variables or templates within the template.
2931          *
2932          * @param array $piece The parts of the template
2933          *  $piece['text']: matched text
2934          *  $piece['title']: the title, i.e. the part before the |
2935          *  $piece['parts']: the parameter array
2936          * @return string the text of the template
2937          * @private
2938          */
2939         function braceSubstitution( $piece ) {
2940                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2941                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2942                 wfProfileIn( $fname );
2943                 wfProfileIn( __METHOD__.'-setup' );
2944
2945                 # Flags
2946                 $found = false;             # $text has been filled
2947                 $nowiki = false;            # wiki markup in $text should be escaped
2948                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2949                 $noargs = false;            # Don't replace triple-brace arguments in $text
2950                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2951                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2952                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2953                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2954
2955                 # Title object, where $text came from
2956                 $title = NULL;
2957
2958                 $linestart = '';
2959
2960
2961                 # $part1 is the bit before the first |, and must contain only title characters
2962                 # $args is a list of arguments, starting from index 0, not including $part1
2963
2964                 $titleText = $part1 = $piece['title'];
2965                 # If the third subpattern matched anything, it will start with |
2966
2967                 if (null == $piece['parts']) {
2968                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2969                         if ($replaceWith != $piece['text']) {
2970                                 $text = $replaceWith;
2971                                 $found = true;
2972                                 $noparse = true;
2973                                 $noargs = true;
2974                         }
2975                 }
2976
2977                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2978                 wfProfileOut( __METHOD__.'-setup' );
2979
2980                 # SUBST
2981                 wfProfileIn( __METHOD__.'-modifiers' );
2982                 if ( !$found ) {
2983                         $mwSubst =& MagicWord::get( 'subst' );
2984                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2985                                 # One of two possibilities is true:
2986                                 # 1) Found SUBST but not in the PST phase
2987                                 # 2) Didn't find SUBST and in the PST phase
2988                                 # In either case, return without further processing
2989                                 $text = $piece['text'];
2990                                 $found = true;
2991                                 $noparse = true;
2992                                 $noargs = true;
2993                         }
2994                 }
2995
2996                 # MSG, MSGNW and RAW
2997                 if ( !$found ) {
2998                         # Check for MSGNW:
2999                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3000                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3001                                 $nowiki = true;
3002                         } else {
3003                                 # Remove obsolete MSG:
3004                                 $mwMsg =& MagicWord::get( 'msg' );
3005                                 $mwMsg->matchStartAndRemove( $part1 );
3006                         }
3007
3008                         # Check for RAW:
3009                         $mwRaw =& MagicWord::get( 'raw' );
3010                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3011                                 $forceRawInterwiki = true;
3012                         }
3013                 }
3014                 wfProfileOut( __METHOD__.'-modifiers' );
3015
3016                 //save path level before recursing into functions & templates.
3017                 $lastPathLevel = $this->mTemplatePath;
3018
3019                 # Parser functions
3020                 if ( !$found ) {
3021                         wfProfileIn( __METHOD__ . '-pfunc' );
3022
3023                         $colonPos = strpos( $part1, ':' );
3024                         if ( $colonPos !== false ) {
3025                                 # Case sensitive functions
3026                                 $function = substr( $part1, 0, $colonPos );
3027                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3028                                         $function = $this->mFunctionSynonyms[1][$function];
3029                                 } else {
3030                                         # Case insensitive functions
3031                                         $function = strtolower( $function );
3032                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3033                                                 $function = $this->mFunctionSynonyms[0][$function];
3034                                         } else {
3035                                                 $function = false;
3036                                         }
3037                                 }
3038                                 if ( $function ) {
3039                                         $funcArgs = array_map( 'trim', $args );
3040                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
3041                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
3042                                         $found = true;
3043
3044                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3045                                         //$noargs = true;
3046                                         //$noparse = true;
3047
3048                                         if ( is_array( $result ) ) {
3049                                                 if ( isset( $result[0] ) ) {
3050                                                         $text = $linestart . $result[0];
3051                                                         unset( $result[0] );
3052                                                 }
3053
3054                                                 // Extract flags into the local scope
3055                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3056                                                 extract( $result );
3057                                         } else {
3058                                                 $text = $linestart . $result;
3059                                         }
3060                                 }
3061                         }
3062                         wfProfileOut( __METHOD__ . '-pfunc' );
3063                 }
3064
3065                 # Template table test
3066
3067                 # Did we encounter this template already? If yes, it is in the cache
3068                 # and we need to check for loops.
3069                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3070                         $found = true;
3071
3072                         # Infinite loop test
3073                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3074                                 $noparse = true;
3075                                 $noargs = true;
3076                                 $found = true;
3077                                 $text = $linestart .
3078                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3079                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3080                         } else {
3081                                 # set $text to cached message.
3082                                 $text = $linestart . $this->mTemplates[$piece['title']];
3083                                 #treat title for cached page the same as others
3084                                 $ns = NS_TEMPLATE;
3085                                 $subpage = '';
3086                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3087                                 if ($subpage !== '') {
3088                                   $ns = $this->mTitle->getNamespace();
3089                                 }
3090                                 $title = Title::newFromText( $part1, $ns );
3091                                 //used by include size checking
3092                                 $titleText = $title->getPrefixedText();
3093                                 //used by edit section links
3094                                 $replaceHeadings = true;
3095
3096                         }
3097                 }
3098
3099                 # Load from database
3100                 if ( !$found ) {
3101                         wfProfileIn( __METHOD__ . '-loadtpl' );
3102                         $ns = NS_TEMPLATE;
3103                         # declaring $subpage directly in the function call
3104                         # does not work correctly with references and breaks
3105                         # {{/subpage}}-style inclusions
3106                         $subpage = '';
3107                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3108                         if ($subpage !== '') {
3109                                 $ns = $this->mTitle->getNamespace();
3110                         }
3111                         $title = Title::newFromText( $part1, $ns );
3112
3113
3114                         if ( !is_null( $title ) ) {
3115                                 $titleText = $title->getPrefixedText();
3116                                 # Check for language variants if the template is not found
3117                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3118                                         $wgContLang->findVariantLink($part1, $title);
3119                                 }
3120
3121                                 if ( !$title->isExternal() ) {
3122                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3123                                                 $text = SpecialPage::capturePath( $title );
3124                                                 if ( is_string( $text ) ) {
3125                                                         $found = true;
3126                                                         $noparse = true;
3127                                                         $noargs = true;
3128                                                         $isHTML = true;
3129                                                         $this->disableCache();
3130                                                 }
3131                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3132                                                 $found = false; //access denied
3133                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3134                                         } else {
3135                                                 list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3136                                                 if ( $articleContent !== false ) {
3137                                                         $found = true;
3138                                                         $text = $articleContent;
3139                                                         $replaceHeadings = true;
3140                                                 }
3141                                         }
3142
3143                                         # If the title is valid but undisplayable, make a link to it
3144                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3145                                                 $text = "[[:$titleText]]";
3146                                                 $found = true;
3147                                         }
3148                                 } elseif ( $title->isTrans() ) {
3149                                         // Interwiki transclusion
3150                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3151                                                 $text = $this->interwikiTransclude( $title, 'render' );
3152                                                 $isHTML = true;
3153                                                 $noparse = true;
3154                                         } else {
3155                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3156                                                 $replaceHeadings = true;
3157                                         }
3158                                         $found = true;
3159                                 }
3160
3161                                 # Template cache array insertion
3162                                 # Use the original $piece['title'] not the mangled $part1, so that
3163                                 # modifiers such as RAW: produce separate cache entries
3164                                 if( $found ) {
3165                                         if( $isHTML ) {
3166                                                 // A special page; don't store it in the template cache.
3167                                         } else {
3168                                                 $this->mTemplates[$piece['title']] = $text;
3169                                         }
3170                                         $text = $linestart . $text;
3171                                 }
3172                         }
3173                         wfProfileOut( __METHOD__ . '-loadtpl' );
3174                 }
3175
3176                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3177                         # Error, oversize inclusion
3178                         $text = $linestart .
3179                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3180                         $noparse = true;
3181                         $noargs = true;
3182                 }
3183
3184                 # Recursive parsing, escaping and link table handling
3185                 # Only for HTML output
3186                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3187                         $text = wfEscapeWikiText( $text );
3188                 } elseif ( !$this->ot['msg'] && $found ) {
3189                         if ( $noargs ) {
3190                                 $assocArgs = array();
3191                         } else {
3192                                 # Clean up argument array
3193                                 $assocArgs = self::createAssocArgs($args);
3194                                 # Add a new element to the templace recursion path
3195                                 $this->mTemplatePath[$part1] = 1;
3196                         }
3197
3198                         if ( !$noparse ) {
3199                                 # If there are any <onlyinclude> tags, only include them
3200                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3201                                         $replacer = new OnlyIncludeReplacer;
3202                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3203                                                 array( &$replacer, 'replace' ), $text );
3204                                         $text = $replacer->output;
3205                                 }
3206                                 # Remove <noinclude> sections and <includeonly> tags
3207                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3208                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3209
3210                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3211                                         # Strip <nowiki>, <pre>, etc.
3212                                         $text = $this->strip( $text, $this->mStripState );
3213                                         if ( $this->ot['html'] ) {
3214                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3215                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3216                                                 $text = Sanitizer::removeHTMLcomments( $text );
3217                                         }
3218                                 }
3219                                 $text = $this->replaceVariables( $text, $assocArgs );
3220
3221                                 # If the template begins with a table or block-level
3222                                 # element, it should be treated as beginning a new line.
3223                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3224                                         $text = "\n" . $text;
3225                                 }
3226                         } elseif ( !$noargs ) {
3227                                 # $noparse and !$noargs
3228                                 # Just replace the arguments, not any double-brace items
3229                                 # This is used for rendered interwiki transclusion
3230                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3231                         }
3232                 }
3233                 # Prune lower levels off the recursion check path
3234                 $this->mTemplatePath = $lastPathLevel;
3235
3236                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3237                         # Error, oversize inclusion
3238                         $text = $linestart .
3239                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3240                         $noparse = true;
3241                         $noargs = true;
3242                 }
3243
3244                 if ( !$found ) {
3245                         wfProfileOut( $fname );
3246                         return $piece['text'];
3247                 } else {
3248                         wfProfileIn( __METHOD__ . '-placeholders' );
3249                         if ( $isHTML ) {
3250                                 # Replace raw HTML by a placeholder
3251                                 # Add a blank line preceding, to prevent it from mucking up
3252                                 # immediately preceding headings
3253                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3254                         } else {
3255                                 # replace ==section headers==
3256                                 # XXX this needs to go away once we have a better parser.
3257                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3258                                         if( !is_null( $title ) )
3259                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3260                                         else
3261                                                 $encodedname = base64_encode("");
3262                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3263                                                 PREG_SPLIT_DELIM_CAPTURE);
3264                                         $text = '';
3265                                         $nsec = $headingOffset;
3266
3267                                         for( $i = 0; $i < count($m); $i += 2 ) {
3268                                                 $text .= $m[$i];
3269                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3270                                                 $hl = $m[$i + 1];
3271                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3272                                                         $text .= $hl;
3273                                                         continue;
3274                                                 }
3275                                                 $m2 = array();
3276                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3277                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3278                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3279
3280                                                 $nsec++;
3281                                         }
3282                                 }
3283                         }
3284                         wfProfileOut( __METHOD__ . '-placeholders' );
3285                 }
3286
3287                 # Prune lower levels off the recursion check path
3288                 $this->mTemplatePath = $lastPathLevel;
3289
3290                 if ( !$found ) {
3291                         wfProfileOut( $fname );
3292                         return $piece['text'];
3293                 } else {
3294                         wfProfileOut( $fname );
3295                         return $text;
3296                 }
3297         }
3298
3299         /**
3300          * Fetch the unparsed text of a template and register a reference to it.
3301          */
3302         function fetchTemplateAndtitle( $title ) {
3303                 $text = $skip = false;
3304                 $finalTitle = $title;
3305                 // Loop to fetch the article, with up to 1 redirect
3306                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3307                         # Give extensions a chance to select the revision instead
3308                         $id = false; // Assume current
3309                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( &$this, &$title, &$skip, &$id ) );
3310
3311                         if( $skip ) {
3312                                 $text = false;
3313                                 $this->mOutput->addTemplate( $title, $title->getArticleID(), null );
3314                                 break;
3315                         }
3316                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3317                         $rev_id = $rev ? $rev->getId() : 0;
3318
3319                         $this->mOutput->addTemplate( $title, $title->getArticleID(), $rev_id );
3320
3321                         if( $rev ) {
3322                                 $text = $rev->getText();
3323                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3324                                 global $wgLang;
3325                                 $message = $wgLang->lcfirst( $title->getText() );
3326                                 $text = wfMsgForContentNoTrans( $message );
3327                                 if( wfEmptyMsg( $message, $text ) ) {
3328                                         $text = false;
3329                                         break;
3330                                 }
3331                         } else {
3332                                 break;
3333                         }
3334                         if ( $text === false ) {
3335                                 break;
3336                         }
3337                         // Redirect?
3338                         $finalTitle = $title;
3339                         $title = Title::newFromRedirect( $text );
3340                 }
3341                 return array($text,$finalTitle);
3342         }
3343
3344         function fetchTemplate( $title ) {
3345                 $rv = $this->fetchTemplateAndtitle($title);
3346                 return $rv[0];
3347         }
3348
3349         /**
3350          * Transclude an interwiki link.
3351          */
3352         function interwikiTransclude( $title, $action ) {
3353                 global $wgEnableScaryTranscluding;
3354
3355                 if (!$wgEnableScaryTranscluding)
3356                         return wfMsg('scarytranscludedisabled');
3357
3358                 $url = $title->getFullUrl( "action=$action" );
3359
3360                 if (strlen($url) > 255)
3361                         return wfMsg('scarytranscludetoolong');
3362                 return $this->fetchScaryTemplateMaybeFromCache($url);
3363         }
3364
3365         function fetchScaryTemplateMaybeFromCache($url) {
3366                 global $wgTranscludeCacheExpiry;
3367                 $dbr = wfGetDB(DB_SLAVE);
3368                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3369                                 array('tc_url' => $url));
3370                 if ($obj) {
3371                         $time = $obj->tc_time;
3372                         $text = $obj->tc_contents;
3373                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3374                                 return $text;
3375                         }
3376                 }
3377
3378                 $text = Http::get($url);
3379                 if (!$text)
3380                         return wfMsg('scarytranscludefailed', $url);
3381
3382                 $dbw = wfGetDB(DB_MASTER);
3383                 $dbw->replace('transcache', array('tc_url'), array(
3384                         'tc_url' => $url,
3385                         'tc_time' => time(),
3386                         'tc_contents' => $text));
3387                 return $text;
3388         }
3389
3390
3391         /**
3392          * Triple brace replacement -- used for template arguments
3393          * @private
3394          */
3395         function argSubstitution( $matches ) {
3396                 $arg = trim( $matches['title'] );
3397                 $text = $matches['text'];
3398                 $inputArgs = end( $this->mArgStack );
3399
3400                 if ( array_key_exists( $arg, $inputArgs ) ) {
3401                         $text = $inputArgs[$arg];
3402                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3403                 null != $matches['parts'] && count($matches['parts']) > 0) {
3404                         $text = $matches['parts'][0];
3405                 }
3406                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3407                         $text = $matches['text'] .
3408                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3409                 }
3410
3411                 return $text;
3412         }
3413
3414         /**
3415          * Increment an include size counter
3416          *
3417          * @param string $type The type of expansion
3418          * @param integer $size The size of the text
3419          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3420          */
3421         function incrementIncludeSize( $type, $size ) {
3422                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3423                         return false;
3424                 } else {
3425                         $this->mIncludeSizes[$type] += $size;
3426                         return true;
3427                 }
3428         }
3429
3430         /**
3431          * Detect __NOGALLERY__ magic word and set a placeholder
3432          */
3433         function stripNoGallery( &$text ) {
3434                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3435                 # do not add TOC
3436                 $mw = MagicWord::get( 'nogallery' );
3437                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3438         }
3439
3440         /**
3441          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3442          * placeholder that will then be replaced by the real TOC in
3443          * ->formatHeadings, this works because at this points real
3444          * comments will have already been discarded by the sanitizer.
3445          *
3446          * Any additional __TOC__ magic words left over will be discarded
3447          * as there can only be one TOC on the page.
3448          */
3449         function stripToc( $text ) {
3450                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3451                 # do not add TOC
3452                 $mw = MagicWord::get( 'notoc' );
3453                 if( $mw->matchAndRemove( $text ) ) {
3454                         $this->mShowToc = false;
3455                 }
3456
3457                 $mw = MagicWord::get( 'toc' );
3458                 if( $mw->match( $text ) ) {
3459                         $this->mShowToc = true;
3460                         $this->mForceTocPosition = true;
3461
3462                         // Set a placeholder. At the end we'll fill it in with the TOC.
3463                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3464
3465                         // Only keep the first one.
3466                         $text = $mw->replace( '', $text );
3467                 }
3468                 return $text;
3469         }
3470
3471         /**
3472          * This function accomplishes several tasks:
3473          * 1) Auto-number headings if that option is enabled
3474          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3475          * 3) Add a Table of contents on the top for users who have enabled the option
3476          * 4) Auto-anchor headings
3477          *
3478          * It loops through all headlines, collects the necessary data, then splits up the
3479          * string and re-inserts the newly formatted headlines.
3480          *
3481          * @param string $text
3482          * @param boolean $isMain
3483          * @private
3484          */
3485         function formatHeadings( $text, $isMain=true ) {
3486                 global $wgMaxTocLevel, $wgContLang;
3487
3488                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3489                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3490                         $showEditLink = 0;
3491                 } else {
3492                         $showEditLink = $this->mOptions->getEditSection();
3493                 }
3494
3495                 # Inhibit editsection links if requested in the page
3496                 $esw =& MagicWord::get( 'noeditsection' );
3497                 if( $esw->matchAndRemove( $text ) ) {
3498                         $showEditLink = 0;
3499                 }
3500
3501                 # Get all headlines for numbering them and adding funky stuff like [edit]
3502                 # links - this is for later, but we need the number of headlines right now
3503                 $matches = array();
3504                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3505
3506                 # if there are fewer than 4 headlines in the article, do not show TOC
3507                 # unless it's been explicitly enabled.
3508                 $enoughToc = $this->mShowToc &&
3509                         (($numMatches >= 4) || $this->mForceTocPosition);
3510
3511                 # Allow user to stipulate that a page should have a "new section"
3512                 # link added via __NEWSECTIONLINK__
3513                 $mw =& MagicWord::get( 'newsectionlink' );
3514                 if( $mw->matchAndRemove( $text ) )
3515                         $this->mOutput->setNewSection( true );
3516
3517                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3518                 # override above conditions and always show TOC above first header
3519                 $mw =& MagicWord::get( 'forcetoc' );
3520                 if ($mw->matchAndRemove( $text ) ) {
3521                         $this->mShowToc = true;
3522                         $enoughToc = true;
3523                 }
3524
3525                 # We need this to perform operations on the HTML
3526                 $sk = $this->mOptions->getSkin();
3527
3528                 # headline counter
3529                 $headlineCount = 0;
3530                 $sectionCount = 0; # headlineCount excluding template sections
3531                 $numVisible = 0;
3532
3533                 # Ugh .. the TOC should have neat indentation levels which can be
3534                 # passed to the skin functions. These are determined here
3535                 $toc = '';
3536                 $full = '';
3537                 $head = array();
3538                 $sublevelCount = array();
3539                 $levelCount = array();
3540                 $toclevel = 0;
3541                 $level = 0;
3542                 $prevlevel = 0;
3543                 $toclevel = 0;
3544                 $prevtoclevel = 0;
3545
3546                 foreach( $matches[3] as $headline ) {
3547                         $istemplate = 0;
3548                         $templatetitle = '';
3549                         $templatesection = 0;
3550                         $numbering = '';
3551                         $mat = array();
3552                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3553                                 $istemplate = 1;
3554                                 $templatetitle = base64_decode($mat[1]);
3555                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3556                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3557                         }
3558
3559                         if( $toclevel ) {
3560                                 $prevlevel = $level;
3561                                 $prevtoclevel = $toclevel;
3562                         }
3563                         $level = $matches[1][$headlineCount];
3564
3565                         if( $doNumberHeadings || $enoughToc ) {
3566
3567                                 if ( $level > $prevlevel ) {
3568                                         # Increase TOC level
3569                                         $toclevel++;
3570                                         $sublevelCount[$toclevel] = 0;
3571                                         if( $toclevel<$wgMaxTocLevel ) {
3572                                                 $prevtoclevel = $toclevel;
3573                                                 $toc .= $sk->tocIndent();
3574                                                 $numVisible++;
3575                                         }
3576                                 }
3577                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3578                                         # Decrease TOC level, find level to jump to
3579
3580                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3581                                                 # Can only go down to level 1
3582                                                 $toclevel = 1;
3583                                         } else {
3584                                                 for ($i = $toclevel; $i > 0; $i--) {
3585                                                         if ( $levelCount[$i] == $level ) {
3586                                                                 # Found last matching level
3587                                                                 $toclevel = $i;
3588                                                                 break;
3589                                                         }
3590                                                         elseif ( $levelCount[$i] < $level ) {
3591                                                                 # Found first matching level below current level
3592                                                                 $toclevel = $i + 1;
3593                                                                 break;
3594                                                         }
3595                                                 }
3596                                         }
3597                                         if( $toclevel<$wgMaxTocLevel ) {
3598                                                 if($prevtoclevel < $wgMaxTocLevel) {
3599                                                         # Unindent only if the previous toc level was shown :p
3600                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3601                                                 } else {
3602                                                         $toc .= $sk->tocLineEnd();
3603                                                 }
3604                                         }
3605                                 }
3606                                 else {
3607                                         # No change in level, end TOC line
3608                                         if( $toclevel<$wgMaxTocLevel ) {
3609                                                 $toc .= $sk->tocLineEnd();
3610                                         }
3611                                 }
3612
3613                                 $levelCount[$toclevel] = $level;
3614
3615                                 # count number of headlines for each level
3616                                 @$sublevelCount[$toclevel]++;
3617                                 $dot = 0;
3618                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3619                                         if( !empty( $sublevelCount[$i] ) ) {
3620                                                 if( $dot ) {
3621                                                         $numbering .= '.';
3622                                                 }
3623                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3624                                                 $dot = 1;
3625                                         }
3626                                 }
3627                         }
3628
3629                         # The canonized header is a version of the header text safe to use for links
3630                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3631                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3632
3633                         # Remove link placeholders by the link text.
3634                         #     <!--LINK number-->
3635                         # turns into
3636                         #     link text with suffix
3637                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3638                                                             "\$this->mLinkHolders['texts'][\$1]",
3639                                                             $canonized_headline );
3640                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3641                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3642                                                             $canonized_headline );
3643
3644                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3645                         $tocline = preg_replace(
3646                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3647                                 array( '',                          '<$1>'),
3648                                 $canonized_headline
3649                         );
3650                         $tocline = trim( $tocline );
3651
3652                         # For the anchor, strip out HTML-y stuff period
3653                         $canonized_headline = preg_replace( '/<.*?'.'>/', '', $canonized_headline );
3654                         $canonized_headline = trim( $canonized_headline );
3655
3656                         # Save headline for section edit hint before it's escaped
3657                         $headline_hint = $canonized_headline;
3658                         $canonized_headline = Sanitizer::escapeId( $canonized_headline );
3659                         $refers[$headlineCount] = $canonized_headline;
3660
3661                         # count how many in assoc. array so we can track dupes in anchors
3662                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3663                         $refcount[$headlineCount]=$refers[$canonized_headline];
3664
3665                         # Don't number the heading if it is the only one (looks silly)
3666                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3667                                 # the two are different if the line contains a link
3668                                 $headline=$numbering . ' ' . $headline;
3669                         }
3670
3671                         # Create the anchor for linking from the TOC to the section
3672                         $anchor = $canonized_headline;
3673                         if($refcount[$headlineCount] > 1 ) {
3674                                 $anchor .= '_' . $refcount[$headlineCount];
3675                         }
3676                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3677                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3678                         }
3679                         # give headline the correct <h#> tag
3680                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3681                                 if( $istemplate )
3682                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3683                                 else
3684                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3685                         } else {
3686                                 $editlink = '';
3687                         }
3688                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3689
3690                         $headlineCount++;
3691                         if( !$istemplate )
3692                                 $sectionCount++;
3693                 }
3694
3695                 # Never ever show TOC if no headers
3696                 if( $numVisible < 1 ) {
3697                         $enoughToc = false;
3698                 }
3699
3700                 if( $enoughToc ) {
3701                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3702                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3703                         }
3704                         $toc = $sk->tocList( $toc );
3705                 }
3706
3707                 # split up and insert constructed headlines
3708
3709                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3710                 $i = 0;
3711
3712                 foreach( $blocks as $block ) {
3713                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3714                                 # This is the [edit] link that appears for the top block of text when
3715                                 # section editing is enabled
3716
3717                                 # Disabled because it broke block formatting
3718                                 # For example, a bullet point in the top line
3719                                 # $full .= $sk->editSectionLink(0);
3720                         }
3721                         $full .= $block;
3722                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3723                                 # Top anchor now in skin
3724                                 $full = $full.$toc;
3725                         }
3726
3727                         if( !empty( $head[$i] ) ) {
3728                                 $full .= $head[$i];
3729                         }
3730                         $i++;
3731                 }
3732                 if( $this->mForceTocPosition ) {
3733                         return str_replace( '<!--MWTOC-->', $toc, $full );
3734                 } else {
3735                         return $full;
3736                 }
3737         }
3738
3739         /**
3740          * Transform wiki markup when saving a page by doing \r\n -> \n
3741          * conversion, substitting signatures, {{subst:}} templates, etc.
3742          *
3743          * @param string $text the text to transform
3744          * @param Title &$title the Title object for the current article
3745          * @param User &$user the User object describing the current user
3746          * @param ParserOptions $options parsing options
3747          * @param bool $clearState whether to clear the parser state first
3748          * @return string the altered wiki markup
3749          * @public
3750          */
3751         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3752                 $this->mOptions = $options;
3753                 $this->mTitle =& $title;
3754                 $this->setOutputType( OT_WIKI );
3755
3756                 if ( $clearState ) {
3757                         $this->clearState();
3758                 }
3759
3760                 $stripState = new StripState;
3761                 $pairs = array(
3762                         "\r\n" => "\n",
3763                 );
3764                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3765                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3766                 $text = $this->pstPass2( $text, $stripState, $user );
3767                 $text = $stripState->unstripBoth( $text );
3768                 return $text;
3769         }
3770
3771         /**
3772          * Pre-save transform helper function
3773          * @private
3774          */
3775         function pstPass2( $text, &$stripState, $user ) {
3776                 global $wgContLang, $wgLocaltimezone;
3777
3778                 /* Note: This is the timestamp saved as hardcoded wikitext to
3779                  * the database, we use $wgContLang here in order to give
3780                  * everyone the same signature and use the default one rather
3781                  * than the one selected in each user's preferences.
3782                  */
3783                 if ( isset( $wgLocaltimezone ) ) {
3784                         $oldtz = getenv( 'TZ' );
3785                         putenv( 'TZ='.$wgLocaltimezone );
3786                 }
3787                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3788                   ' (' . date( 'T' ) . ')';
3789                 if ( isset( $wgLocaltimezone ) ) {
3790                         putenv( 'TZ='.$oldtz );
3791                 }
3792
3793                 # Variable replacement
3794                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3795                 $text = $this->replaceVariables( $text );
3796
3797                 # Strip out <nowiki> etc. added via replaceVariables
3798                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3799
3800                 # Signatures
3801                 $sigText = $this->getUserSig( $user );
3802                 $text = strtr( $text, array(
3803                         '~~~~~' => $d,
3804                         '~~~~' => "$sigText $d",
3805                         '~~~' => $sigText
3806                 ) );
3807
3808                 # Context links: [[|name]] and [[name (context)|]]
3809                 #
3810                 global $wgLegalTitleChars;
3811                 $tc = "[$wgLegalTitleChars]";
3812                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3813
3814                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3815                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3816                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3817
3818                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3819                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3820                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3821
3822                 $t = $this->mTitle->getText();
3823                 $m = array();
3824                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3825                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3826                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3827                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3828                 } else {
3829                         # if there's no context, don't bother duplicating the title
3830                         $text = preg_replace( $p2, '[[\\1]]', $text );
3831                 }
3832
3833                 # Trim trailing whitespace
3834                 $text = rtrim( $text );
3835
3836                 return $text;
3837         }
3838
3839         /**
3840          * Fetch the user's signature text, if any, and normalize to
3841          * validated, ready-to-insert wikitext.
3842          *
3843          * @param User $user
3844          * @return string
3845          * @private
3846          */
3847         function getUserSig( &$user ) {
3848                 global $wgMaxSigChars;
3849
3850                 $username = $user->getName();
3851                 $nickname = $user->getOption( 'nickname' );
3852                 $nickname = $nickname === '' ? $username : $nickname;
3853
3854                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3855                         $nickname = $username;
3856                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3857                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3858                         # Sig. might contain markup; validate this
3859                         if( $this->validateSig( $nickname ) !== false ) {
3860                                 # Validated; clean up (if needed) and return it
3861                                 return $this->cleanSig( $nickname, true );
3862                         } else {
3863                                 # Failed to validate; fall back to the default
3864                                 $nickname = $username;
3865                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3866                         }
3867                 }
3868
3869                 // Make sure nickname doesnt get a sig in a sig
3870                 $nickname = $this->cleanSigInSig( $nickname );
3871
3872                 # If we're still here, make it a link to the user page
3873                 $userpage = $user->getUserPage();
3874                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3875         }
3876
3877         /**
3878          * Check that the user's signature contains no bad XML
3879          *
3880          * @param string $text
3881          * @return mixed An expanded string, or false if invalid.
3882          */
3883         function validateSig( $text ) {
3884                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3885         }
3886
3887         /**
3888          * Clean up signature text
3889          *
3890          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3891          * 2) Substitute all transclusions
3892          *
3893          * @param string $text
3894          * @param $parsing Whether we're cleaning (preferences save) or parsing
3895          * @return string Signature text
3896          */
3897         function cleanSig( $text, $parsing = false ) {
3898                 global $wgTitle;
3899                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3900
3901                 $substWord = MagicWord::get( 'subst' );
3902                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3903                 $substText = '{{' . $substWord->getSynonym( 0 );
3904
3905                 $text = preg_replace( $substRegex, $substText, $text );
3906                 $text = $this->cleanSigInSig( $text );
3907                 $text = $this->replaceVariables( $text );
3908
3909                 $this->clearState();
3910                 return $text;
3911         }
3912
3913         /**
3914          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3915          * @param string $text
3916          * @return string Signature text with /~{3,5}/ removed
3917          */
3918         function cleanSigInSig( $text ) {
3919                 $text = preg_replace( '/~{3,5}/', '', $text );
3920                 return $text;
3921         }
3922
3923         /**
3924          * Set up some variables which are usually set up in parse()
3925          * so that an external function can call some class members with confidence
3926          * @public
3927          */
3928         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3929                 $this->mTitle =& $title;
3930                 $this->mOptions = $options;
3931                 $this->setOutputType( $outputType );
3932                 if ( $clearState ) {
3933                         $this->clearState();
3934                 }
3935         }
3936
3937         /**
3938          * Transform a MediaWiki message by replacing magic variables.
3939          *
3940          * @param string $text the text to transform
3941          * @param ParserOptions $options  options
3942          * @return string the text with variables substituted
3943          * @public
3944          */
3945         function transformMsg( $text, $options ) {
3946                 global $wgTitle;
3947                 static $executing = false;
3948
3949                 $fname = "Parser::transformMsg";
3950
3951                 # Guard against infinite recursion
3952                 if ( $executing ) {
3953                         return $text;
3954                 }
3955                 $executing = true;
3956
3957                 wfProfileIn($fname);
3958
3959                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3960                         $this->mTitle = $wgTitle;
3961                 } else {
3962                         $this->mTitle = Title::newFromText('msg');
3963                 }
3964                 $this->mOptions = $options;
3965                 $this->setOutputType( OT_MSG );
3966                 $this->clearState();
3967                 $text = $this->replaceVariables( $text );
3968
3969                 $executing = false;
3970                 wfProfileOut($fname);
3971                 return $text;
3972         }
3973
3974         /**
3975          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3976          * The callback should have the following form:
3977          *    function myParserHook( $text, $params, &$parser ) { ... }
3978          *
3979          * Transform and return $text. Use $parser for any required context, e.g. use
3980          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3981          *
3982          * @public
3983          *
3984          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3985          * @param mixed $callback The callback function (and object) to use for the tag
3986          *
3987          * @return The old value of the mTagHooks array associated with the hook
3988          */
3989         function setHook( $tag, $callback ) {
3990                 $tag = strtolower( $tag );
3991                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3992                 $this->mTagHooks[$tag] = $callback;
3993
3994                 return $oldVal;
3995         }
3996
3997         function setTransparentTagHook( $tag, $callback ) {
3998                 $tag = strtolower( $tag );
3999                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4000                 $this->mTransparentTagHooks[$tag] = $callback;
4001
4002                 return $oldVal;
4003         }
4004
4005         /**
4006          * Create a function, e.g. {{sum:1|2|3}}
4007          * The callback function should have the form:
4008          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4009          *
4010          * The callback may either return the text result of the function, or an array with the text
4011          * in element 0, and a number of flags in the other elements. The names of the flags are
4012          * specified in the keys. Valid flags are:
4013          *   found                     The text returned is valid, stop processing the template. This
4014          *                             is on by default.
4015          *   nowiki                    Wiki markup in the return value should be escaped
4016          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4017          *   noargs                    Don't replace triple-brace arguments in the return value
4018          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4019          *
4020          * @public
4021          *
4022          * @param string $id The magic word ID
4023          * @param mixed $callback The callback function (and object) to use
4024          * @param integer $flags a combination of the following flags:
4025          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4026          *
4027          * @return The old callback function for this name, if any
4028          */
4029         function setFunctionHook( $id, $callback, $flags = 0 ) {
4030                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
4031                 $this->mFunctionHooks[$id] = $callback;
4032
4033                 # Add to function cache
4034                 $mw = MagicWord::get( $id );
4035                 if( !$mw )
4036                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4037
4038                 $synonyms = $mw->getSynonyms();
4039                 $sensitive = intval( $mw->isCaseSensitive() );
4040
4041                 foreach ( $synonyms as $syn ) {
4042                         # Case
4043                         if ( !$sensitive ) {
4044                                 $syn = strtolower( $syn );
4045                         }
4046                         # Add leading hash
4047                         if ( !( $flags & SFH_NO_HASH ) ) {
4048                                 $syn = '#' . $syn;
4049                         }
4050                         # Remove trailing colon
4051                         if ( substr( $syn, -1, 1 ) == ':' ) {
4052                                 $syn = substr( $syn, 0, -1 );
4053                         }
4054                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4055                 }
4056                 return $oldVal;
4057         }
4058
4059         /**
4060          * Get all registered function hook identifiers
4061          *
4062          * @return array
4063          */
4064         function getFunctionHooks() {
4065                 return array_keys( $this->mFunctionHooks );
4066         }
4067
4068         /**
4069          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4070          * Placeholders created in Skin::makeLinkObj()
4071          * Returns an array of links found, indexed by PDBK:
4072          *  0 - broken
4073          *  1 - normal link
4074          *  2 - stub
4075          * $options is a bit field, RLH_FOR_UPDATE to select for update
4076          */
4077         function replaceLinkHolders( &$text, $options = 0 ) {
4078                 global $wgUser;
4079                 global $wgContLang;
4080
4081                 $fname = 'Parser::replaceLinkHolders';
4082                 wfProfileIn( $fname );
4083
4084                 $pdbks = array();
4085                 $colours = array();
4086                 $sk = $this->mOptions->getSkin();
4087                 $linkCache =& LinkCache::singleton();
4088
4089                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4090                         wfProfileIn( $fname.'-check' );
4091                         $dbr = wfGetDB( DB_SLAVE );
4092                         $page = $dbr->tableName( 'page' );
4093                         $threshold = $wgUser->getOption('stubthreshold');
4094
4095                         # Sort by namespace
4096                         asort( $this->mLinkHolders['namespaces'] );
4097
4098                         # Generate query
4099                         $query = false;
4100                         $current = null;
4101                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4102                                 # Make title object
4103                                 $title = $this->mLinkHolders['titles'][$key];
4104
4105                                 # Skip invalid entries.
4106                                 # Result will be ugly, but prevents crash.
4107                                 if ( is_null( $title ) ) {
4108                                         continue;
4109                                 }
4110                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4111
4112                                 # Check if it's a static known link, e.g. interwiki
4113                                 if ( $title->isAlwaysKnown() ) {
4114                                         $colours[$pdbk] = 1;
4115                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4116                                         $colours[$pdbk] = 1;
4117                                         $this->mOutput->addLink( $title, $id );
4118                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4119                                         $colours[$pdbk] = 0;
4120                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4121                                         $colours[$pdbk] = 0;
4122                                 } else {
4123                                         # Not in the link cache, add it to the query
4124                                         if ( !isset( $current ) ) {
4125                                                 $current = $ns;
4126                                                 $query =  "SELECT page_id, page_namespace, page_title";
4127                                                 if ( $threshold > 0 ) {
4128                                                         $query .= ', page_len, page_is_redirect';
4129                                                 }
4130                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4131                                         } elseif ( $current != $ns ) {
4132                                                 $current = $ns;
4133                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4134                                         } else {
4135                                                 $query .= ', ';
4136                                         }
4137
4138                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4139                                 }
4140                         }
4141                         if ( $query ) {
4142                                 $query .= '))';
4143                                 if ( $options & RLH_FOR_UPDATE ) {
4144                                         $query .= ' FOR UPDATE';
4145                                 }
4146
4147                                 $res = $dbr->query( $query, $fname );
4148
4149                                 # Fetch data and form into an associative array
4150                                 # non-existent = broken
4151                                 # 1 = known
4152                                 # 2 = stub
4153                                 while ( $s = $dbr->fetchObject($res) ) {
4154                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4155                                         $pdbk = $title->getPrefixedDBkey();
4156                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4157                                         $this->mOutput->addLink( $title, $s->page_id );
4158
4159                                         $colours[$pdbk] = ( $threshold == 0 || (
4160                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4161                                                                 $s->page_is_redirect ||
4162                                                                 !Namespace::isContent( $s->page_namespace ) )
4163                                                             ? 1 : 2 );
4164                                 }
4165                         }
4166                         wfProfileOut( $fname.'-check' );
4167
4168                         # Do a second query for different language variants of links and categories
4169                         if($wgContLang->hasVariants()){
4170                                 $linkBatch = new LinkBatch();
4171                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4172                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4173                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4174
4175                                 $categories = $this->mOutput->getCategoryLinks();
4176
4177                                 // Add variants of links to link batch
4178                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4179                                         $title = $this->mLinkHolders['titles'][$key];
4180                                         if ( is_null( $title ) )
4181                                                 continue;
4182
4183                                         $pdbk = $title->getPrefixedDBkey();
4184                                         $titleText = $title->getText();
4185
4186                                         // generate all variants of the link title text
4187                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4188
4189                                         // if link was not found (in first query), add all variants to query
4190                                         if ( !isset($colours[$pdbk]) ){
4191                                                 foreach($allTextVariants as $textVariant){
4192                                                         if($textVariant != $titleText){
4193                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4194                                                                 if(is_null($variantTitle)) continue;
4195                                                                 $linkBatch->addObj( $variantTitle );
4196                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4197                                                         }
4198                                                 }
4199                                         }
4200                                 }
4201
4202                                 // process categories, check if a category exists in some variant
4203                                 foreach( $categories as $category ){
4204                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4205                                         foreach($variants as $variant){
4206                                                 if($variant != $category){
4207                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4208                                                         if(is_null($variantTitle)) continue;
4209                                                         $linkBatch->addObj( $variantTitle );
4210                                                         $categoryMap[$variant] = $category;
4211                                                 }
4212                                         }
4213                                 }
4214
4215
4216                                 if(!$linkBatch->isEmpty()){
4217                                         // construct query
4218                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4219
4220                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4221                                         if ( $threshold > 0 ) {
4222                                                 $variantQuery .= ', page_len, page_is_redirect';
4223                                         }
4224
4225                                         $variantQuery .= " FROM $page WHERE $titleClause";
4226                                         if ( $options & RLH_FOR_UPDATE ) {
4227                                                 $variantQuery .= ' FOR UPDATE';
4228                                         }
4229
4230                                         $varRes = $dbr->query( $variantQuery, $fname );
4231
4232                                         // for each found variants, figure out link holders and replace
4233                                         while ( $s = $dbr->fetchObject($varRes) ) {
4234
4235                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4236                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4237                                                 $vardbk = $variantTitle->getDBkey();
4238
4239                                                 $holderKeys = array();
4240                                                 if(isset($variantMap[$varPdbk])){
4241                                                         $holderKeys = $variantMap[$varPdbk];
4242                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4243                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4244                                                 }
4245
4246                                                 // loop over link holders
4247                                                 foreach($holderKeys as $key){
4248                                                         $title = $this->mLinkHolders['titles'][$key];
4249                                                         if ( is_null( $title ) ) continue;
4250
4251                                                         $pdbk = $title->getPrefixedDBkey();
4252
4253                                                         if(!isset($colours[$pdbk])){
4254                                                                 // found link in some of the variants, replace the link holder data
4255                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4256                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4257
4258                                                                 // set pdbk and colour
4259                                                                 $pdbks[$key] = $varPdbk;
4260                                                                 if ( $threshold >  0 ) {
4261                                                                         $size = $s->page_len;
4262                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4263                                                                                 $colours[$varPdbk] = 1;
4264                                                                         } else {
4265                                                                                 $colours[$varPdbk] = 2;
4266                                                                         }
4267                                                                 }
4268                                                                 else {
4269                                                                         $colours[$varPdbk] = 1;
4270                                                                 }
4271                                                         }
4272                                                 }
4273
4274                                                 // check if the object is a variant of a category
4275                                                 if(isset($categoryMap[$vardbk])){
4276                                                         $oldkey = $categoryMap[$vardbk];
4277                                                         if($oldkey != $vardbk)
4278                                                                 $varCategories[$oldkey]=$vardbk;
4279                                                 }
4280                                         }
4281
4282                                         // rebuild the categories in original order (if there are replacements)
4283                                         if(count($varCategories)>0){
4284                                                 $newCats = array();
4285                                                 $originalCats = $this->mOutput->getCategories();
4286                                                 foreach($originalCats as $cat => $sortkey){
4287                                                         // make the replacement
4288                                                         if( array_key_exists($cat,$varCategories) )
4289                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4290                                                         else $newCats[$cat] = $sortkey;
4291                                                 }
4292                                                 $this->mOutput->setCategoryLinks($newCats);
4293                                         }
4294                                 }
4295                         }
4296
4297                         # Construct search and replace arrays
4298                         wfProfileIn( $fname.'-construct' );
4299                         $replacePairs = array();
4300                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4301                                 $pdbk = $pdbks[$key];
4302                                 $searchkey = "<!--LINK $key-->";
4303                                 $title = $this->mLinkHolders['titles'][$key];
4304                                 if ( empty( $colours[$pdbk] ) ) {
4305                                         $linkCache->addBadLinkObj( $title );
4306                                         $colours[$pdbk] = 0;
4307                                         $this->mOutput->addLink( $title, 0 );
4308                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4309                                                                         $this->mLinkHolders['texts'][$key],
4310                                                                         $this->mLinkHolders['queries'][$key] );
4311                                 } elseif ( $colours[$pdbk] == 1 ) {
4312                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4313                                                                         $this->mLinkHolders['texts'][$key],
4314                                                                         $this->mLinkHolders['queries'][$key] );
4315                                 } elseif ( $colours[$pdbk] == 2 ) {
4316                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4317                                                                         $this->mLinkHolders['texts'][$key],
4318                                                                         $this->mLinkHolders['queries'][$key] );
4319                                 }
4320                         }
4321                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4322                         wfProfileOut( $fname.'-construct' );
4323
4324                         # Do the thing
4325                         wfProfileIn( $fname.'-replace' );
4326                         $text = preg_replace_callback(
4327                                 '/(<!--LINK .*?-->)/',
4328                                 $replacer->cb(),
4329                                 $text);
4330
4331                         wfProfileOut( $fname.'-replace' );
4332                 }
4333
4334                 # Now process interwiki link holders
4335                 # This is quite a bit simpler than internal links
4336                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4337                         wfProfileIn( $fname.'-interwiki' );
4338                         # Make interwiki link HTML
4339                         $replacePairs = array();
4340                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4341                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4342                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4343                         }
4344                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4345
4346                         $text = preg_replace_callback(
4347                                 '/<!--IWLINK (.*?)-->/',
4348                                 $replacer->cb(),
4349                                 $text );
4350                         wfProfileOut( $fname.'-interwiki' );
4351                 }
4352
4353                 wfProfileOut( $fname );
4354                 return $colours;
4355         }
4356
4357         /**
4358          * Replace <!--LINK--> link placeholders with plain text of links
4359          * (not HTML-formatted).
4360          * @param string $text
4361          * @return string
4362          */
4363         function replaceLinkHoldersText( $text ) {
4364                 $fname = 'Parser::replaceLinkHoldersText';
4365                 wfProfileIn( $fname );
4366
4367                 $text = preg_replace_callback(
4368                         '/<!--(LINK|IWLINK) (.*?)-->/',
4369                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4370                         $text );
4371
4372                 wfProfileOut( $fname );
4373                 return $text;
4374         }
4375
4376         /**
4377          * @param array $matches
4378          * @return string
4379          * @private
4380          */
4381         function replaceLinkHoldersTextCallback( $matches ) {
4382                 $type = $matches[1];
4383                 $key  = $matches[2];
4384                 if( $type == 'LINK' ) {
4385                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4386                                 return $this->mLinkHolders['texts'][$key];
4387                         }
4388                 } elseif( $type == 'IWLINK' ) {
4389                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4390                                 return $this->mInterwikiLinkHolders['texts'][$key];
4391                         }
4392                 }
4393                 return $matches[0];
4394         }
4395
4396         /**
4397          * Tag hook handler for 'pre'.
4398          */
4399         function renderPreTag( $text, $attribs ) {
4400                 // Backwards-compatibility hack
4401                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4402
4403                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4404                 return wfOpenElement( 'pre', $attribs ) .
4405                         Xml::escapeTagsOnly( $content ) .
4406                         '</pre>';
4407         }
4408
4409         /**
4410          * Renders an image gallery from a text with one line per image.
4411          * text labels may be given by using |-style alternative text. E.g.
4412          *   Image:one.jpg|The number "1"
4413          *   Image:tree.jpg|A tree
4414          * given as text will return the HTML of a gallery with two images,
4415          * labeled 'The number "1"' and
4416          * 'A tree'.
4417          */
4418         function renderImageGallery( $text, $params ) {
4419                 $ig = new ImageGallery();
4420                 $ig->setContextTitle( $this->mTitle );
4421                 $ig->setShowBytes( false );
4422                 $ig->setShowFilename( false );
4423                 $ig->setParser( $this );
4424                 $ig->setHideBadImages();
4425                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4426                 $ig->useSkin( $this->mOptions->getSkin() );
4427                 $ig->mRevisionId = $this->mRevisionId;
4428
4429                 if( isset( $params['caption'] ) ) {
4430                         $caption = $params['caption'];
4431                         $caption = htmlspecialchars( $caption );
4432                         $caption = $this->replaceInternalLinks( $caption );
4433                         $ig->setCaptionHtml( $caption );
4434                 }
4435                 if( isset( $params['perrow'] ) ) {
4436                         $ig->setPerRow( $params['perrow'] );
4437                 }
4438                 if( isset( $params['widths'] ) ) {
4439                         $ig->setWidths( $params['widths'] );
4440                 }
4441                 if( isset( $params['heights'] ) ) {
4442                         $ig->setHeights( $params['heights'] );
4443                 }
4444
4445                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4446
4447                 $lines = explode( "\n", $text );
4448                 foreach ( $lines as $line ) {
4449                         # match lines like these:
4450                         # Image:someimage.jpg|This is some image
4451                         $matches = array();
4452                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4453                         # Skip empty lines
4454                         if ( count( $matches ) == 0 ) {
4455                                 continue;
4456                         }
4457                         $tp = Title::newFromText( $matches[1] );
4458                         $nt =& $tp;
4459                         if( is_null( $nt ) ) {
4460                                 # Bogus title. Ignore these so we don't bomb out later.
4461                                 continue;
4462                         }
4463                         if ( isset( $matches[3] ) ) {
4464                                 $label = $matches[3];
4465                         } else {
4466                                 $label = '';
4467                         }
4468
4469                         $pout = $this->parse( $label,
4470                                 $this->mTitle,
4471                                 $this->mOptions,
4472                                 false, // Strip whitespace...?
4473                                 false  // Don't clear state!
4474                         );
4475                         $html = $pout->getText();
4476
4477                         $ig->add( $nt, $html );
4478
4479                         # Only add real images (bug #5586)
4480                         if ( $nt->getNamespace() == NS_IMAGE ) {
4481                                 $this->mOutput->addImage( $nt->getDBkey() );
4482                         }
4483                 }
4484                 return $ig->toHTML();
4485         }
4486
4487         function getImageParams( $handler ) {
4488                 if ( $handler ) {
4489                         $handlerClass = get_class( $handler );
4490                 } else {
4491                         $handlerClass = '';
4492                 }
4493                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4494                         // Initialise static lists
4495                         static $internalParamNames = array(
4496                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4497                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4498                                         'bottom', 'text-bottom' ),
4499                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4500                                         'upright', 'border' ),
4501                         );
4502                         static $internalParamMap;
4503                         if ( !$internalParamMap ) {
4504                                 $internalParamMap = array();
4505                                 foreach ( $internalParamNames as $type => $names ) {
4506                                         foreach ( $names as $name ) {
4507                                                 $magicName = str_replace( '-', '_', "img_$name" );
4508                                                 $internalParamMap[$magicName] = array( $type, $name );
4509                                         }
4510                                 }
4511                         }
4512
4513                         // Add handler params
4514                         $paramMap = $internalParamMap;
4515                         if ( $handler ) {
4516                                 $handlerParamMap = $handler->getParamMap();
4517                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4518                                         $paramMap[$magic] = array( 'handler', $paramName );
4519                                 }
4520                         }
4521                         $this->mImageParams[$handlerClass] = $paramMap;
4522                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4523                 }
4524                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4525         }
4526
4527         /**
4528          * Parse image options text and use it to make an image
4529          */
4530         function makeImage( $title, $options ) {
4531                 # @TODO: let the MediaHandler specify its transform parameters
4532                 #
4533                 # Check if the options text is of the form "options|alt text"
4534                 # Options are:
4535                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4536                 #  * left               no resizing, just left align. label is used for alt= only
4537                 #  * right              same, but right aligned
4538                 #  * none               same, but not aligned
4539                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4540                 #  * center             center the image
4541                 #  * framed             Keep original image size, no magnify-button.
4542                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4543                 #  * upright            reduce width for upright images, rounded to full __0 px
4544                 #  * border             draw a 1px border around the image
4545                 # vertical-align values (no % or length right now):
4546                 #  * baseline
4547                 #  * sub
4548                 #  * super
4549                 #  * top
4550                 #  * text-top
4551                 #  * middle
4552                 #  * bottom
4553                 #  * text-bottom
4554
4555                 $parts = array_map( 'trim', explode( '|', $options) );
4556                 $sk = $this->mOptions->getSkin();
4557
4558                 # Give extensions a chance to select the file revision for us
4559                 $skip = $time = false;
4560                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4561
4562                 if ( $skip ) {
4563                         return $sk->makeLinkObj( $title );
4564                 }
4565
4566                 # Get parameter map
4567                 $file = wfFindFile( $title, $time );
4568                 $handler = $file ? $file->getHandler() : false;
4569
4570                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4571
4572                 # Process the input parameters
4573                 $caption = '';
4574                 $params = array( 'frame' => array(), 'handler' => array(),
4575                         'horizAlign' => array(), 'vertAlign' => array() );
4576                 foreach( $parts as $part ) {
4577                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4578                         if ( isset( $paramMap[$magicName] ) ) {
4579                                 list( $type, $paramName ) = $paramMap[$magicName];
4580                                 $params[$type][$paramName] = $value;
4581
4582                                 // Special case; width and height come in one variable together
4583                                 if( $type == 'handler' && $paramName == 'width' ) {
4584                                         $m = array();
4585                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4586                                                 $params[$type]['width'] = intval( $m[1] );
4587                                                 $params[$type]['height'] = intval( $m[2] );
4588                                         } else {
4589                                                 $params[$type]['width'] = intval( $value );
4590                                         }
4591                                 }
4592                         } else {
4593                                 $caption = $part;
4594                         }
4595                 }
4596
4597                 # Process alignment parameters
4598                 if ( $params['horizAlign'] ) {
4599                         $params['frame']['align'] = key( $params['horizAlign'] );
4600                 }
4601                 if ( $params['vertAlign'] ) {
4602                         $params['frame']['valign'] = key( $params['vertAlign'] );
4603                 }
4604
4605                 # Validate the handler parameters
4606                 if ( $handler ) {
4607                         foreach ( $params['handler'] as $name => $value ) {
4608                                 if ( !$handler->validateParam( $name, $value ) ) {
4609                                         unset( $params['handler'][$name] );
4610                                 }
4611                         }
4612                 }
4613
4614                 # Strip bad stuff out of the alt text
4615                 $alt = $this->replaceLinkHoldersText( $caption );
4616
4617                 # make sure there are no placeholders in thumbnail attributes
4618                 # that are later expanded to html- so expand them now and
4619                 # remove the tags
4620                 $alt = $this->mStripState->unstripBoth( $alt );
4621                 $alt = Sanitizer::stripAllTags( $alt );
4622
4623                 $params['frame']['alt'] = $alt;
4624                 $params['frame']['caption'] = $caption;
4625
4626                 # Linker does the rest
4627                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4628
4629                 # Give the handler a chance to modify the parser object
4630                 if ( $handler ) {
4631                         $handler->parserTransformHook( $this, $file );
4632                 }
4633
4634                 return $ret;
4635         }
4636
4637         /**
4638          * Set a flag in the output object indicating that the content is dynamic and
4639          * shouldn't be cached.
4640          */
4641         function disableCache() {
4642                 wfDebug( "Parser output marked as uncacheable.\n" );
4643                 $this->mOutput->mCacheTime = -1;
4644         }
4645
4646         /**#@+
4647          * Callback from the Sanitizer for expanding items found in HTML attribute
4648          * values, so they can be safely tested and escaped.
4649          * @param string $text
4650          * @param array $args
4651          * @return string
4652          * @private
4653          */
4654         function attributeStripCallback( &$text, $args ) {
4655                 $text = $this->replaceVariables( $text, $args );
4656                 $text = $this->mStripState->unstripBoth( $text );
4657                 return $text;
4658         }
4659
4660         /**#@-*/
4661
4662         /**#@+
4663          * Accessor/mutator
4664          */
4665         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4666         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4667         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4668         /**#@-*/
4669
4670         /**#@+
4671          * Accessor
4672          */
4673         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4674         /**#@-*/
4675
4676
4677         /**
4678          * Break wikitext input into sections, and either pull or replace
4679          * some particular section's text.
4680          *
4681          * External callers should use the getSection and replaceSection methods.
4682          *
4683          * @param $text Page wikitext
4684          * @param $section Numbered section. 0 pulls the text before the first
4685          *                 heading; other numbers will pull the given section
4686          *                 along with its lower-level subsections.
4687          * @param $mode One of "get" or "replace"
4688          * @param $newtext Replacement text for section data.
4689          * @return string for "get", the extracted section text.
4690          *                for "replace", the whole page with the section replaced.
4691          */
4692         private function extractSections( $text, $section, $mode, $newtext='' ) {
4693                 # I.... _hope_ this is right.
4694                 # Otherwise, sometimes we don't have things initialized properly.
4695                 $this->clearState();
4696
4697                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4698                 # comments to be stripped as well)
4699                 $stripState = new StripState;
4700
4701                 $oldOutputType = $this->mOutputType;
4702                 $oldOptions = $this->mOptions;
4703                 $this->mOptions = new ParserOptions();
4704                 $this->setOutputType( OT_WIKI );
4705
4706                 $striptext = $this->strip( $text, $stripState, true );
4707
4708                 $this->setOutputType( $oldOutputType );
4709                 $this->mOptions = $oldOptions;
4710
4711                 # now that we can be sure that no pseudo-sections are in the source,
4712                 # split it up by section
4713                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4714                 $comment = "(?:$uniq-!--.*?QINU\x07)";
4715                 $secs = preg_split(
4716                         "/
4717                         (
4718                                 ^
4719                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4720                                 (=+) # Should this be limited to 6?
4721                                 .+?  # Section title...
4722                                 \\2  # Ending = count must match start
4723                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4724                                 $
4725                         |
4726                                 <h([1-6])\b.*?>
4727                                 .*?
4728                                 <\/h\\3\s*>
4729                         )
4730                         /mix",
4731                         $striptext, -1,
4732                         PREG_SPLIT_DELIM_CAPTURE);
4733
4734                 if( $mode == "get" ) {
4735                         if( $section == 0 ) {
4736                                 // "Section 0" returns the content before any other section.
4737                                 $rv = $secs[0];
4738                         } else {
4739                                 //track missing section, will replace if found.
4740                                 $rv = $newtext;
4741                         }
4742                 } elseif( $mode == "replace" ) {
4743                         if( $section == 0 ) {
4744                                 $rv = $newtext . "\n\n";
4745                                 $remainder = true;
4746                         } else {
4747                                 $rv = $secs[0];
4748                                 $remainder = false;
4749                         }
4750                 }
4751                 $count = 0;
4752                 $sectionLevel = 0;
4753                 for( $index = 1; $index < count( $secs ); ) {
4754                         $headerLine = $secs[$index++];
4755                         if( $secs[$index] ) {
4756                                 // A wiki header
4757                                 $headerLevel = strlen( $secs[$index++] );
4758                         } else {
4759                                 // An HTML header
4760                                 $index++;
4761                                 $headerLevel = intval( $secs[$index++] );
4762                         }
4763                         $content = $secs[$index++];
4764
4765                         $count++;
4766                         if( $mode == "get" ) {
4767                                 if( $count == $section ) {
4768                                         $rv = $headerLine . $content;
4769                                         $sectionLevel = $headerLevel;
4770                                 } elseif( $count > $section ) {
4771                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4772                                                 $rv .= $headerLine . $content;
4773                                         } else {
4774                                                 // Broke out to a higher-level section
4775                                                 break;
4776                                         }
4777                                 }
4778                         } elseif( $mode == "replace" ) {
4779                                 if( $count < $section ) {
4780                                         $rv .= $headerLine . $content;
4781                                 } elseif( $count == $section ) {
4782                                         $rv .= $newtext . "\n\n";
4783                                         $sectionLevel = $headerLevel;
4784                                 } elseif( $count > $section ) {
4785                                         if( $headerLevel <= $sectionLevel ) {
4786                                                 // Passed the section's sub-parts.
4787                                                 $remainder = true;
4788                                         }
4789                                         if( $remainder ) {
4790                                                 $rv .= $headerLine . $content;
4791                                         }
4792                                 }
4793                         }
4794                 }
4795                 if (is_string($rv))
4796                         # reinsert stripped tags
4797                         $rv = trim( $stripState->unstripBoth( $rv ) );
4798
4799                 return $rv;
4800         }
4801
4802         /**
4803          * This function returns the text of a section, specified by a number ($section).
4804          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4805          * the first section before any such heading (section 0).
4806          *
4807          * If a section contains subsections, these are also returned.
4808          *
4809          * @param $text String: text to look in
4810          * @param $section Integer: section number
4811          * @param $deftext: default to return if section is not found
4812          * @return string text of the requested section
4813          */
4814         public function getSection( $text, $section, $deftext='' ) {
4815                 return $this->extractSections( $text, $section, "get", $deftext );
4816         }
4817
4818         public function replaceSection( $oldtext, $section, $text ) {
4819                 return $this->extractSections( $oldtext, $section, "replace", $text );
4820         }
4821
4822         /**
4823          * Get the timestamp associated with the current revision, adjusted for
4824          * the default server-local timestamp
4825          */
4826         function getRevisionTimestamp() {
4827                 if ( is_null( $this->mRevisionTimestamp ) ) {
4828                         wfProfileIn( __METHOD__ );
4829                         global $wgContLang;
4830                         $dbr = wfGetDB( DB_SLAVE );
4831                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4832                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4833
4834                         // Normalize timestamp to internal MW format for timezone processing.
4835                         // This has the added side-effect of replacing a null value with
4836                         // the current time, which gives us more sensible behavior for
4837                         // previews.
4838                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4839
4840                         // The cryptic '' timezone parameter tells to use the site-default
4841                         // timezone offset instead of the user settings.
4842                         //
4843                         // Since this value will be saved into the parser cache, served
4844                         // to other users, and potentially even used inside links and such,
4845                         // it needs to be consistent for all visitors.
4846                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4847
4848                         wfProfileOut( __METHOD__ );
4849                 }
4850                 return $this->mRevisionTimestamp;
4851         }
4852
4853         /**
4854          * Mutator for $mDefaultSort
4855          *
4856          * @param $sort New value
4857          */
4858         public function setDefaultSort( $sort ) {
4859                 $this->mDefaultSort = $sort;
4860         }
4861
4862         /**
4863          * Accessor for $mDefaultSort
4864          * Will use the title/prefixed title if none is set
4865          *
4866          * @return string
4867          */
4868         public function getDefaultSort() {
4869                 if( $this->mDefaultSort !== false ) {
4870                         return $this->mDefaultSort;
4871                 } else {
4872                         return $this->mTitle->getNamespace() == NS_CATEGORY
4873                                         ? $this->mTitle->getText()
4874                                         : $this->mTitle->getPrefixedText();
4875                 }
4876         }
4877
4878         /**
4879          * Try to guess the section anchor name based on a wikitext fragment
4880          * presumably extracted from a heading, for example "Header" from
4881          * "== Header ==".
4882          */
4883         public function guessSectionNameFromWikiText( $text ) {
4884                 # Strip out wikitext links(they break the anchor)
4885                 $text = $this->stripSectionName( $text );
4886                 $headline = Sanitizer::decodeCharReferences( $text );
4887                 # strip out HTML
4888                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4889                 $headline = trim( $headline );
4890                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4891                 $replacearray = array(
4892                         '%3A' => ':',
4893                         '%' => '.'
4894                 );
4895                 return str_replace(
4896                         array_keys( $replacearray ),
4897                         array_values( $replacearray ),
4898                         $sectionanchor );
4899         }
4900
4901         /**
4902          * Strips a text string of wikitext for use in a section anchor
4903          *
4904          * Accepts a text string and then removes all wikitext from the
4905          * string and leaves only the resultant text (i.e. the result of
4906          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4907          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4908          * to create valid section anchors by mimicing the output of the
4909          * parser when headings are parsed.
4910          *
4911          * @param $text string Text string to be stripped of wikitext
4912          * for use in a Section anchor
4913          * @return Filtered text string
4914          */
4915         public function stripSectionName( $text ) {
4916                 # Strip internal link markup
4917                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4918                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4919
4920                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4921                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4922                 # on how many empty links there are on the page - need to figure that out.
4923                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4924
4925                 # Parse wikitext quotes (italics & bold)
4926                 $text = $this->doQuotes($text);
4927
4928                 # Strip HTML tags
4929                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4930                 return $text;
4931         }
4932 }
4933
4934 /**
4935  * @todo document, briefly.
4936  * @addtogroup Parser
4937  */
4938 class OnlyIncludeReplacer {
4939         var $output = '';
4940
4941         function replace( $matches ) {
4942                 if ( substr( $matches[1], -1 ) == "\n" ) {
4943                         $this->output .= substr( $matches[1], 0, -1 );
4944                 } else {
4945                         $this->output .= $matches[1];
4946                 }
4947         }
4948 }
4949
4950 /**
4951  * @todo document, briefly.
4952  * @addtogroup Parser
4953  */
4954 class StripState {
4955         var $general, $nowiki;
4956
4957         function __construct() {
4958                 $this->general = new ReplacementArray;
4959                 $this->nowiki = new ReplacementArray;
4960         }
4961
4962         function unstripGeneral( $text ) {
4963                 wfProfileIn( __METHOD__ );
4964                 $text = $this->general->replace( $text );
4965                 wfProfileOut( __METHOD__ );
4966                 return $text;
4967         }
4968
4969         function unstripNoWiki( $text ) {
4970                 wfProfileIn( __METHOD__ );
4971                 $text = $this->nowiki->replace( $text );
4972                 wfProfileOut( __METHOD__ );
4973                 return $text;
4974         }
4975
4976         function unstripBoth( $text ) {
4977                 wfProfileIn( __METHOD__ );
4978                 $text = $this->general->replace( $text );
4979                 $text = $this->nowiki->replace( $text );
4980                 wfProfileOut( __METHOD__ );
4981                 return $text;
4982         }
4983 }