includes/Parser_OldPP.php

   1 <?php
   2 /**
   3  * Parser with old preprocessor
   4  */
   5 class Parser_OldPP
   6 {
   7         /**
   8          * Update this version number when the ParserOutput format
   9          * changes in an incompatible way, so the parser cache
  10          * can automatically discard old data.
  11          */
  12         const VERSION = '1.6.4';
  13
  14         # Flags for Parser::setFunctionHook
  15         # Also available as global constants from Defines.php
  16         const SFH_NO_HASH = 1;
  17
  18         # Constants needed for external link processing
  19         # Everything except bracket, space, or control characters
  20         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  21         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)\\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/S';
  22
  23         // State constants for the definition list colon extraction
  24         const COLON_STATE_TEXT = 0;
  25         const COLON_STATE_TAG = 1;
  26         const COLON_STATE_TAGSTART = 2;
  27         const COLON_STATE_CLOSETAG = 3;
  28         const COLON_STATE_TAGSLASH = 4;
  29         const COLON_STATE_COMMENT = 5;
  30         const COLON_STATE_COMMENTDASH = 6;
  31         const COLON_STATE_COMMENTDASHDASH = 7;
  32
  33         /**#@+
  34          * @private
  35          */
  36         # Persistent:
  37         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  38                 $mImageParams, $mImageParamsMagicArray, $mExtLinkBracketedRegex;
  39
  40         # Cleared with clearState():
  41         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  42         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  43         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
  44         var $mIncludeSizes, $mDefaultSort;
  45         var $mTemplates,        // cache of already loaded templates, avoids
  46                                 // multiple SQL queries for the same string
  47             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  48                                 // in this path. Used for loop detection.
  49
  50         # Temporary
  51         # These are variables reset at least once per parse regardless of $clearState
  52         var $mOptions,      // ParserOptions object
  53                 $mTitle,        // Title context, used for self-link rendering and similar things
  54                 $mOutputType,   // Output type, one of the OT_xxx constants
  55                 $ot,            // Shortcut alias, see setOutputType()
  56                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  57                 $mRevisionTimestamp, // The timestamp of the specified revision ID
  58                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
  59
  60         /**#@-*/
  61
  62         /**
  63          * Constructor
  64          *
  65          * @public
  66          */
  67         function __construct( $conf = array() ) {
  68                 $this->mTagHooks = array();
  69                 $this->mTransparentTagHooks = array();
  70                 $this->mFunctionHooks = array();
  71                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
  72                 $this->mFirstCall = true;
  73                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
  74                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
  75         }
  76
  77         /**
  78          * Do various kinds of initialisation on the first call of the parser
  79          */
  80         function firstCallInit() {
  81                 if ( !$this->mFirstCall ) {
  82                         return;
  83                 }
  84
  85                 wfProfileIn( __METHOD__ );
  86                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
  87
  88                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
  89
  90                 # Syntax for arguments (see self::setFunctionHook):
  91                 #  "name for lookup in localized magic words array",
  92                 #  function callback,
  93                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
  94                 #    instead of {{#int:...}})
  95                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
  96                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
  97                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
  98                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
  99                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 100                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 101                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 102                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 103                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 104                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 105                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 106                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 107                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 108                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 109                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 110                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 111                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 112                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 113                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 114                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 115                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 116                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 117                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 118                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 119                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 120                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 121                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 122
 123                 if ( $wgAllowDisplayTitle ) {
 124                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 125                 }
 126                 if ( $wgAllowSlowParserFunctions ) {
 127                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 128                 }
 129
 130                 $this->initialiseVariables();
 131                 $this->mFirstCall = false;
 132                 wfProfileOut( __METHOD__ );
 133         }
 134
 135         /**
 136          * Clear Parser state
 137          *
 138          * @private
 139          */
 140         function clearState() {
 141                 wfProfileIn( __METHOD__ );
 142                 if ( $this->mFirstCall ) {
 143                         $this->firstCallInit();
 144                 }
 145                 $this->mOutput = new ParserOutput;
 146                 $this->mAutonumber = 0;
 147                 $this->mLastSection = '';
 148                 $this->mDTopen = false;
 149                 $this->mIncludeCount = array();
 150                 $this->mStripState = new StripState;
 151                 $this->mArgStack = array();
 152                 $this->mInPre = false;
 153                 $this->mInterwikiLinkHolders = array(
 154                         'texts' => array(),
 155                         'titles' => array()
 156                 );
 157                 $this->mLinkHolders = array(
 158                         'namespaces' => array(),
 159                         'dbkeys' => array(),
 160                         'queries' => array(),
 161                         'texts' => array(),
 162                         'titles' => array()
 163                 );
 164                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 165
 166                 /**
 167                  * Prefix for temporary replacement strings for the multipass parser.
 168                  * \x07 should never appear in input as it's disallowed in XML.
 169                  * Using it at the front also gives us a little extra robustness
 170                  * since it shouldn't match when butted up against identifier-like
 171                  * string constructs.
 172                  */
 173                 $this->mUniqPrefix = "\x07UNIQ" . self::getRandomString();
 174
 175                 # Clear these on every parse, bug 4549
 176                 $this->mTemplates = array();
 177                 $this->mTemplatePath = array();
 178
 179                 $this->mShowToc = true;
 180                 $this->mForceTocPosition = false;
 181                 $this->mIncludeSizes = array(
 182                         'pre-expand' => 0,
 183                         'post-expand' => 0,
 184                         'arg' => 0
 185                 );
 186                 $this->mDefaultSort = false;
 187
 188                 wfRunHooks( 'ParserClearState', array( &$this ) );
 189                 wfProfileOut( __METHOD__ );
 190         }
 191
 192         function setOutputType( $ot ) {
 193                 $this->mOutputType = $ot;
 194                 // Shortcut alias
 195                 $this->ot = array(
 196                         'html' => $ot == OT_HTML,
 197                         'wiki' => $ot == OT_WIKI,
 198                         'msg' => $ot == OT_MSG,
 199                         'pre' => $ot == OT_PREPROCESS,
 200                 );
 201         }
 202
 203         /**
 204          * Accessor for mUniqPrefix.
 205          *
 206          * @public
 207          */
 208         function uniqPrefix() {
 209                 return $this->mUniqPrefix;
 210         }
 211
 212         /**
 213          * Convert wikitext to HTML
 214          * Do not call this function recursively.
 215          *
 216          * @param string $text Text we want to parse
 217          * @param Title &$title A title object
 218          * @param array $options
 219          * @param boolean $linestart
 220          * @param boolean $clearState
 221          * @param int $revid number to pass in {{REVISIONID}}
 222          * @return ParserOutput a ParserOutput
 223          */
 224         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 225                 /**
 226                  * First pass--just handle <nowiki> sections, pass the rest off
 227                  * to internalParse() which does all the real work.
 228                  */
 229
 230                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 231                 $fname = 'Parser::parse-' . wfGetCaller();
 232                 wfProfileIn( __METHOD__ );
 233                 wfProfileIn( $fname );
 234
 235                 if ( $clearState ) {
 236                         $this->clearState();
 237                 }
 238
 239                 $this->mOptions = $options;
 240                 $this->mTitle =& $title;
 241                 $oldRevisionId = $this->mRevisionId;
 242                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 243                 if( $revid !== null ) {
 244                         $this->mRevisionId = $revid;
 245                         $this->mRevisionTimestamp = null;
 246                 }
 247                 $this->setOutputType( OT_HTML );
 248                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 249                 $text = $this->strip( $text, $this->mStripState );
 250                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 251                 $text = $this->internalParse( $text );
 252                 $text = $this->mStripState->unstripGeneral( $text );
 253
 254                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 255                 $fixtags = array(
 256                         # french spaces, last one Guillemet-left
 257                         # only if there is something before the space
 258                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 259                         # french spaces, Guillemet-right
 260                         '/(\\302\\253) /' => '\\1&nbsp;',
 261                 );
 262                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 263
 264                 # only once and last
 265                 $text = $this->doBlockLevels( $text, $linestart );
 266
 267                 $this->replaceLinkHolders( $text );
 268
 269                 # the position of the parserConvert() call should not be changed. it
 270                 # assumes that the links are all replaced and the only thing left
 271                 # is the <nowiki> mark.
 272                 # Side-effects: this calls $this->mOutput->setTitleText()
 273                 $text = $wgContLang->parserConvert( $text, $this );
 274
 275                 $text = $this->mStripState->unstripNoWiki( $text );
 276
 277                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 278
 279 //!JF Move to its own function
 280
 281                 $uniq_prefix = $this->mUniqPrefix;
 282                 $matches = array();
 283                 $elements = array_keys( $this->mTransparentTagHooks );
 284                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 285
 286                 foreach( $matches as $marker => $data ) {
 287                         list( $element, $content, $params, $tag ) = $data;
 288                         $tagName = strtolower( $element );
 289                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 290                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 291                                         array( $content, $params, $this ) );
 292                         } else {
 293                                 $output = $tag;
 294                         }
 295                         $this->mStripState->general->setPair( $marker, $output );
 296                 }
 297                 $text = $this->mStripState->unstripGeneral( $text );
 298
 299                 $text = Sanitizer::normalizeCharReferences( $text );
 300
 301                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 302                         $text = self::tidy($text);
 303                 } else {
 304                         # attempt to sanitize at least some nesting problems
 305                         # (bug #2702 and quite a few others)
 306                         $tidyregs = array(
 307                                 # ''Something [http://www.cool.com cool''] -->
 308                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 309                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 310                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 311                                 # fix up an anchor inside another anchor, only
 312                                 # at least for a single single nested link (bug 3695)
 313                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 314                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 315                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 316                                 # contains a div, so fix it up here; replace
 317                                 # div with escaped text
 318                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 319                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 320                                 # remove empty italic or bold tag pairs, some
 321                                 # introduced by rules above
 322                                 '/<([bi])><\/\\1>/' => '',
 323                         );
 324
 325                         $text = preg_replace(
 326                                 array_keys( $tidyregs ),
 327                                 array_values( $tidyregs ),
 328                                 $text );
 329                 }
 330
 331                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 332
 333                 # Information on include size limits, for the benefit of users who try to skirt them
 334                 if ( $this->mOptions->getEnableLimitReport() ) {
 335                         $max = $this->mOptions->getMaxIncludeSize();
 336                         $limitReport =
 337                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']}/$max bytes\n" .
 338                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 339                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 340                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 341                         $text .= "<!-- \n$limitReport-->\n";
 342                 }
 343                 $this->mOutput->setText( $text );
 344                 $this->mRevisionId = $oldRevisionId;
 345                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 346                 wfProfileOut( $fname );
 347                 wfProfileOut( __METHOD__ );
 348
 349                 return $this->mOutput;
 350         }
 351
 352         /**
 353          * Recursive parser entry point that can be called from an extension tag
 354          * hook.
 355          */
 356         function recursiveTagParse( $text ) {
 357                 wfProfileIn( __METHOD__ );
 358                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 359                 $text = $this->strip( $text, $this->mStripState );
 360                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 361                 $text = $this->internalParse( $text );
 362                 wfProfileOut( __METHOD__ );
 363                 return $text;
 364         }
 365
 366         /**
 367          * Expand templates and variables in the text, producing valid, static wikitext.
 368          * Also removes comments.
 369          */
 370         function preprocess( $text, $title, $options, $revid = null ) {
 371                 wfProfileIn( __METHOD__ );
 372                 $this->clearState();
 373                 $this->setOutputType( OT_PREPROCESS );
 374                 $this->mOptions = $options;
 375                 $this->mTitle = $title;
 376                 if( $revid !== null ) {
 377                         $this->mRevisionId = $revid;
 378                 }
 379                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 380                 $text = $this->strip( $text, $this->mStripState );
 381                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 382                 if ( $this->mOptions->getRemoveComments() ) {
 383                         $text = Sanitizer::removeHTMLcomments( $text );
 384                 }
 385                 $text = $this->replaceVariables( $text );
 386                 $text = $this->mStripState->unstripBoth( $text );
 387                 wfProfileOut( __METHOD__ );
 388                 return $text;
 389         }
 390
 391         /**
 392          * Get a random string
 393          *
 394          * @private
 395          * @static
 396          */
 397         function getRandomString() {
 398                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 399         }
 400
 401         function &getTitle() { return $this->mTitle; }
 402         function getOptions() { return $this->mOptions; }
 403
 404         function getFunctionLang() {
 405                 global $wgLang, $wgContLang;
 406                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 407         }
 408
 409         /**
 410          * Replaces all occurrences of HTML-style comments and the given tags
 411          * in the text with a random marker and returns teh next text. The output
 412          * parameter $matches will be an associative array filled with data in
 413          * the form:
 414          *   'UNIQ-xxxxx' => array(
 415          *     'element',
 416          *     'tag content',
 417          *     array( 'param' => 'x' ),
 418          *     '<element param="x">tag content</element>' ) )
 419          *
 420          * @param $elements list of element names. Comments are always extracted.
 421          * @param $text Source text string.
 422          * @param $uniq_prefix
 423          *
 424          * @public
 425          * @static
 426          */
 427         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 428                 static $n = 1;
 429                 $stripped = '';
 430                 $matches = array();
 431
 432                 $taglist = implode( '|', $elements );
 433                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 434
 435                 while ( '' != $text ) {
 436                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 437                         $stripped .= $p[0];
 438                         if( count( $p ) < 5 ) {
 439                                 break;
 440                         }
 441                         if( count( $p ) > 5 ) {
 442                                 // comment
 443                                 $element    = $p[4];
 444                                 $attributes = '';
 445                                 $close      = '';
 446                                 $inside     = $p[5];
 447                         } else {
 448                                 // tag
 449                                 $element    = $p[1];
 450                                 $attributes = $p[2];
 451                                 $close      = $p[3];
 452                                 $inside     = $p[4];
 453                         }
 454
 455                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 456                         $stripped .= $marker;
 457
 458                         if ( $close === '/>' ) {
 459                                 // Empty element tag, <tag />
 460                                 $content = null;
 461                                 $text = $inside;
 462                                 $tail = null;
 463                         } else {
 464                                 if( $element == '!--' ) {
 465                                         $end = '/(-->)/';
 466                                 } else {
 467                                         $end = "/(<\\/$element\\s*>)/i";
 468                                 }
 469                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 470                                 $content = $q[0];
 471                                 if( count( $q ) < 3 ) {
 472                                         # No end tag -- let it run out to the end of the text.
 473                                         $tail = '';
 474                                         $text = '';
 475                                 } else {
 476                                         $tail = $q[1];
 477                                         $text = $q[2];
 478                                 }
 479                         }
 480
 481                         $matches[$marker] = array( $element,
 482                                 $content,
 483                                 Sanitizer::decodeTagAttributes( $attributes ),
 484                                 "<$element$attributes$close$content$tail" );
 485                 }
 486                 return $stripped;
 487         }
 488
 489         /**
 490          * Strips and renders nowiki, pre, math, hiero
 491          * If $render is set, performs necessary rendering operations on plugins
 492          * Returns the text, and fills an array with data needed in unstrip()
 493          *
 494          * @param StripState $state
 495          *
 496          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 497          *  will be stripped in addition to other tags. This is important
 498          *  for section editing, where these comments cause confusion when
 499          *  counting the sections in the wikisource
 500          *
 501          * @param array dontstrip contains tags which should not be stripped;
 502          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 503          *
 504          * @private
 505          */
 506         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 507                 global $wgContLang;
 508                 wfProfileIn( __METHOD__ );
 509                 $render = ($this->mOutputType == OT_HTML);
 510
 511                 $uniq_prefix = $this->mUniqPrefix;
 512                 $commentState = new ReplacementArray;
 513                 $nowikiItems = array();
 514                 $generalItems = array();
 515
 516                 $elements = array_merge(
 517                         array( 'nowiki', 'gallery' ),
 518                         array_keys( $this->mTagHooks ) );
 519                 global $wgRawHtml;
 520                 if( $wgRawHtml ) {
 521                         $elements[] = 'html';
 522                 }
 523                 if( $this->mOptions->getUseTeX() ) {
 524                         $elements[] = 'math';
 525                 }
 526
 527                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 528                 foreach ( $elements AS $k => $v ) {
 529                         if ( !in_array ( $v , $dontstrip ) ) continue;
 530                         unset ( $elements[$k] );
 531                 }
 532
 533                 $matches = array();
 534                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 535
 536                 foreach( $matches as $marker => $data ) {
 537                         list( $element, $content, $params, $tag ) = $data;
 538                         if( $render ) {
 539                                 $tagName = strtolower( $element );
 540                                 wfProfileIn( __METHOD__."-render-$tagName" );
 541                                 switch( $tagName ) {
 542                                 case '!--':
 543                                         // Comment
 544                                         if( substr( $tag, -3 ) == '-->' ) {
 545                                                 $output = $tag;
 546                                         } else {
 547                                                 // Unclosed comment in input.
 548                                                 // Close it so later stripping can remove it
 549                                                 $output = "$tag-->";
 550                                         }
 551                                         break;
 552                                 case 'html':
 553                                         if( $wgRawHtml ) {
 554                                                 $output = $content;
 555                                                 break;
 556                                         }
 557                                         // Shouldn't happen otherwise. :)
 558                                 case 'nowiki':
 559                                         $output = Xml::escapeTagsOnly( $content );
 560                                         break;
 561                                 case 'math':
 562                                         $output = $wgContLang->armourMath(
 563                                                 MathRenderer::renderMath( $content, $params ) );
 564                                         break;
 565                                 case 'gallery':
 566                                         $output = $this->renderImageGallery( $content, $params );
 567                                         break;
 568                                 default:
 569                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 570                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 571                                                         array( $content, $params, $this ) );
 572                                         } else {
 573                                                 throw new MWException( "Invalid call hook $element" );
 574                                         }
 575                                 }
 576                                 wfProfileOut( __METHOD__."-render-$tagName" );
 577                         } else {
 578                                 // Just stripping tags; keep the source
 579                                 $output = $tag;
 580                         }
 581
 582                         // Unstrip the output, to support recursive strip() calls
 583                         $output = $state->unstripBoth( $output );
 584
 585                         if( !$stripcomments && $element == '!--' ) {
 586                                 $commentState->setPair( $marker, $output );
 587                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 588                                 $nowikiItems[$marker] = $output;
 589                         } else {
 590                                 $generalItems[$marker] = $output;
 591                         }
 592                 }
 593                 # Add the new items to the state
 594                 # We do this after the loop instead of during it to avoid slowing
 595                 # down the recursive unstrip
 596                 $state->nowiki->mergeArray( $nowikiItems );
 597                 $state->general->mergeArray( $generalItems );
 598
 599                 # Unstrip comments unless explicitly told otherwise.
 600                 # (The comments are always stripped prior to this point, so as to
 601                 # not invoke any extension tags / parser hooks contained within
 602                 # a comment.)
 603                 if ( !$stripcomments ) {
 604                         // Put them all back and forget them
 605                         $text = $commentState->replace( $text );
 606                 }
 607
 608                 wfProfileOut( __METHOD__ );
 609                 return $text;
 610         }
 611
 612         /**
 613          * Restores pre, math, and other extensions removed by strip()
 614          *
 615          * always call unstripNoWiki() after this one
 616          * @private
 617          * @deprecated use $this->mStripState->unstrip()
 618          */
 619         function unstrip( $text, $state ) {
 620                 return $state->unstripGeneral( $text );
 621         }
 622
 623         /**
 624          * Always call this after unstrip() to preserve the order
 625          *
 626          * @private
 627          * @deprecated use $this->mStripState->unstrip()
 628          */
 629         function unstripNoWiki( $text, $state ) {
 630                 return $state->unstripNoWiki( $text );
 631         }
 632
 633         /**
 634          * @deprecated use $this->mStripState->unstripBoth()
 635          */
 636         function unstripForHTML( $text ) {
 637                 return $this->mStripState->unstripBoth( $text );
 638         }
 639
 640         /**
 641          * Add an item to the strip state
 642          * Returns the unique tag which must be inserted into the stripped text
 643          * The tag will be replaced with the original text in unstrip()
 644          *
 645          * @private
 646          */
 647         function insertStripItem( $text, &$state ) {
 648                 $rnd = $this->mUniqPrefix . '-item' . self::getRandomString();
 649                 $state->general->setPair( $rnd, $text );
 650                 return $rnd;
 651         }
 652
 653         /**
 654          * Interface with html tidy, used if $wgUseTidy = true.
 655          * If tidy isn't able to correct the markup, the original will be
 656          * returned in all its glory with a warning comment appended.
 657          *
 658          * Either the external tidy program or the in-process tidy extension
 659          * will be used depending on availability. Override the default
 660          * $wgTidyInternal setting to disable the internal if it's not working.
 661          *
 662          * @param string $text Hideous HTML input
 663          * @return string Corrected HTML output
 664          * @public
 665          * @static
 666          */
 667         function tidy( $text ) {
 668                 global $wgTidyInternal;
 669                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 670 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 671 '<head><title>test</title></head><body>'.$text.'</body></html>';
 672                 if( $wgTidyInternal ) {
 673                         $correctedtext = self::internalTidy( $wrappedtext );
 674                 } else {
 675                         $correctedtext = self::externalTidy( $wrappedtext );
 676                 }
 677                 if( is_null( $correctedtext ) ) {
 678                         wfDebug( "Tidy error detected!\n" );
 679                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 680                 }
 681                 return $correctedtext;
 682         }
 683
 684         /**
 685          * Spawn an external HTML tidy process and get corrected markup back from it.
 686          *
 687          * @private
 688          * @static
 689          */
 690         function externalTidy( $text ) {
 691                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 692                 $fname = 'Parser::externalTidy';
 693                 wfProfileIn( $fname );
 694
 695                 $cleansource = '';
 696                 $opts = ' -utf8';
 697
 698                 $descriptorspec = array(
 699                         0 => array('pipe', 'r'),
 700                         1 => array('pipe', 'w'),
 701                         2 => array('file', wfGetNull(), 'a')
 702                 );
 703                 $pipes = array();
 704                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 705                 if (is_resource($process)) {
 706                         // Theoretically, this style of communication could cause a deadlock
 707                         // here. If the stdout buffer fills up, then writes to stdin could
 708                         // block. This doesn't appear to happen with tidy, because tidy only
 709                         // writes to stdout after it's finished reading from stdin. Search
 710                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 711                         fwrite($pipes[0], $text);
 712                         fclose($pipes[0]);
 713                         while (!feof($pipes[1])) {
 714                                 $cleansource .= fgets($pipes[1], 1024);
 715                         }
 716                         fclose($pipes[1]);
 717                         proc_close($process);
 718                 }
 719
 720                 wfProfileOut( $fname );
 721
 722                 if( $cleansource == '' && $text != '') {
 723                         // Some kind of error happened, so we couldn't get the corrected text.
 724                         // Just give up; we'll use the source text and append a warning.
 725                         return null;
 726                 } else {
 727                         return $cleansource;
 728                 }
 729         }
 730
 731         /**
 732          * Use the HTML tidy PECL extension to use the tidy library in-process,
 733          * saving the overhead of spawning a new process.
 734          *
 735          * 'pear install tidy' should be able to compile the extension module.
 736          *
 737          * @private
 738          * @static
 739          */
 740         function internalTidy( $text ) {
 741                 global $wgTidyConf, $IP;
 742                 $fname = 'Parser::internalTidy';
 743                 wfProfileIn( $fname );
 744
 745                 $tidy = new tidy;
 746                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 747                 $tidy->cleanRepair();
 748                 if( $tidy->getStatus() == 2 ) {
 749                         // 2 is magic number for fatal error
 750                         // http://www.php.net/manual/en/function.tidy-get-status.php
 751                         $cleansource = null;
 752                 } else {
 753                         $cleansource = tidy_get_output( $tidy );
 754                 }
 755                 wfProfileOut( $fname );
 756                 return $cleansource;
 757         }
 758
 759         /**
 760          * parse the wiki syntax used to render tables
 761          *
 762          * @private
 763          */
 764         function doTableStuff ( $text ) {
 765                 $fname = 'Parser::doTableStuff';
 766                 wfProfileIn( $fname );
 767
 768                 $lines = explode ( "\n" , $text );
 769                 $td_history = array (); // Is currently a td tag open?
 770                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 771                 $tr_history = array (); // Is currently a tr tag open?
 772                 $tr_attributes = array (); // history of tr attributes
 773                 $has_opened_tr = array(); // Did this table open a <tr> element?
 774                 $indent_level = 0; // indent level of the table
 775                 foreach ( $lines as $key => $line )
 776                 {
 777                         $line = trim ( $line );
 778
 779                         if( $line == '' ) { // empty line, go to next line
 780                                 continue;
 781                         }
 782                         $first_character = $line{0};
 783                         $matches = array();
 784
 785                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 786                                 // First check if we are starting a new table
 787                                 $indent_level = strlen( $matches[1] );
 788
 789                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 790                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 791
 792                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 793                                 array_push ( $td_history , false );
 794                                 array_push ( $last_tag_history , '' );
 795                                 array_push ( $tr_history , false );
 796                                 array_push ( $tr_attributes , '' );
 797                                 array_push ( $has_opened_tr , false );
 798                         } else if ( count ( $td_history ) == 0 ) {
 799                                 // Don't do any of the following
 800                                 continue;
 801                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 802                                 // We are ending a table
 803                                 $line = '</table>' . substr ( $line , 2 );
 804                                 $last_tag = array_pop ( $last_tag_history );
 805
 806                                 if ( !array_pop ( $has_opened_tr ) ) {
 807                                         $line = "<tr><td></td></tr>{$line}";
 808                                 }
 809
 810                                 if ( array_pop ( $tr_history ) ) {
 811                                         $line = "</tr>{$line}";
 812                                 }
 813
 814                                 if ( array_pop ( $td_history ) ) {
 815                                         $line = "</{$last_tag}>{$line}";
 816                                 }
 817                                 array_pop ( $tr_attributes );
 818                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 819                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 820                                 // Now we have a table row
 821                                 $line = preg_replace( '#^\|-+#', '', $line );
 822
 823                                 // Whats after the tag is now only attributes
 824                                 $attributes = $this->mStripState->unstripBoth( $line );
 825                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 826                                 array_pop ( $tr_attributes );
 827                                 array_push ( $tr_attributes , $attributes );
 828
 829                                 $line = '';
 830                                 $last_tag = array_pop ( $last_tag_history );
 831                                 array_pop ( $has_opened_tr );
 832                                 array_push ( $has_opened_tr , true );
 833
 834                                 if ( array_pop ( $tr_history ) ) {
 835                                         $line = '</tr>';
 836                                 }
 837
 838                                 if ( array_pop ( $td_history ) ) {
 839                                         $line = "</{$last_tag}>{$line}";
 840                                 }
 841
 842                                 $lines[$key] = $line;
 843                                 array_push ( $tr_history , false );
 844                                 array_push ( $td_history , false );
 845                                 array_push ( $last_tag_history , '' );
 846                         }
 847                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 848                                 // This might be cell elements, td, th or captions
 849                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 850                                         $first_character = '+';
 851                                         $line = substr ( $line , 1 );
 852                                 }
 853
 854                                 $line = substr ( $line , 1 );
 855
 856                                 if ( $first_character == '!' ) {
 857                                         $line = str_replace ( '!!' , '||' , $line );
 858                                 }
 859
 860                                 // Split up multiple cells on the same line.
 861                                 // FIXME : This can result in improper nesting of tags processed
 862                                 // by earlier parser steps, but should avoid splitting up eg
 863                                 // attribute values containing literal "||".
 864                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 865
 866                                 $lines[$key] = '';
 867
 868                                 // Loop through each table cell
 869                                 foreach ( $cells as $cell )
 870                                 {
 871                                         $previous = '';
 872                                         if ( $first_character != '+' )
 873                                         {
 874                                                 $tr_after = array_pop ( $tr_attributes );
 875                                                 if ( !array_pop ( $tr_history ) ) {
 876                                                         $previous = "<tr{$tr_after}>\n";
 877                                                 }
 878                                                 array_push ( $tr_history , true );
 879                                                 array_push ( $tr_attributes , '' );
 880                                                 array_pop ( $has_opened_tr );
 881                                                 array_push ( $has_opened_tr , true );
 882                                         }
 883
 884                                         $last_tag = array_pop ( $last_tag_history );
 885
 886                                         if ( array_pop ( $td_history ) ) {
 887                                                 $previous = "</{$last_tag}>{$previous}";
 888                                         }
 889
 890                                         if ( $first_character == '|' ) {
 891                                                 $last_tag = 'td';
 892                                         } else if ( $first_character == '!' ) {
 893                                                 $last_tag = 'th';
 894                                         } else if ( $first_character == '+' ) {
 895                                                 $last_tag = 'caption';
 896                                         } else {
 897                                                 $last_tag = '';
 898                                         }
 899
 900                                         array_push ( $last_tag_history , $last_tag );
 901
 902                                         // A cell could contain both parameters and data
 903                                         $cell_data = explode ( '|' , $cell , 2 );
 904
 905                                         // Bug 553: Note that a '|' inside an invalid link should not
 906                                         // be mistaken as delimiting cell parameters
 907                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 908                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 909                                         } else if ( count ( $cell_data ) == 1 )
 910                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 911                                         else {
 912                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 913                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 914                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 915                                         }
 916
 917                                         $lines[$key] .= $cell;
 918                                         array_push ( $td_history , true );
 919                                 }
 920                         }
 921                 }
 922
 923                 // Closing open td, tr && table
 924                 while ( count ( $td_history ) > 0 )
 925                 {
 926                         if ( array_pop ( $td_history ) ) {
 927                                 $lines[] = '</td>' ;
 928                         }
 929                         if ( array_pop ( $tr_history ) ) {
 930                                 $lines[] = '</tr>' ;
 931                         }
 932                         if ( !array_pop ( $has_opened_tr ) ) {
 933                                 $lines[] = "<tr><td></td></tr>" ;
 934                         }
 935
 936                         $lines[] = '</table>' ;
 937                 }
 938
 939                 $output = implode ( "\n" , $lines ) ;
 940
 941                 // special case: don't return empty table
 942                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 943                         $output = '';
 944                 }
 945
 946                 wfProfileOut( $fname );
 947
 948                 return $output;
 949         }
 950
 951         /**
 952          * Helper function for parse() that transforms wiki markup into
 953          * HTML. Only called for $mOutputType == OT_HTML.
 954          *
 955          * @private
 956          */
 957         function internalParse( $text ) {
 958                 $args = array();
 959                 $isMain = true;
 960                 $fname = 'Parser::internalParse';
 961                 wfProfileIn( $fname );
 962
 963                 # Hook to suspend the parser in this state
 964                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 965                         wfProfileOut( $fname );
 966                         return $text ;
 967                 }
 968
 969                 # Remove <noinclude> tags and <includeonly> sections
 970                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 971                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 972                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 973
 974                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
 975
 976                 $text = $this->replaceVariables( $text, $args );
 977                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 978
 979                 // Tables need to come after variable replacement for things to work
 980                 // properly; putting them before other transformations should keep
 981                 // exciting things like link expansions from showing up in surprising
 982                 // places.
 983                 $text = $this->doTableStuff( $text );
 984
 985                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 986
 987                 $text = $this->stripToc( $text );
 988                 $this->stripNoGallery( $text );
 989                 $text = $this->doHeadings( $text );
 990                 if($this->mOptions->getUseDynamicDates()) {
 991                         $df =& DateFormatter::getInstance();
 992                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 993                 }
 994                 $text = $this->doAllQuotes( $text );
 995                 $text = $this->replaceInternalLinks( $text );
 996                 $text = $this->replaceExternalLinks( $text );
 997
 998                 # replaceInternalLinks may sometimes leave behind
 999                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1000                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1001
1002                 $text = $this->doMagicLinks( $text );
1003                 $text = $this->formatHeadings( $text, $isMain );
1004
1005                 wfProfileOut( $fname );
1006                 return $text;
1007         }
1008
1009         /**
1010          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1011          * magic external links.
1012          *
1013          * @private
1014          */
1015         function &doMagicLinks( &$text ) {
1016                 wfProfileIn( __METHOD__ );
1017                 $text = preg_replace_callback(
1018                         '!(?:                           # Start cases
1019                             <a.*?</a> |                 # Skip link text
1020                             <.*?> |                     # Skip stuff inside HTML elements
1021                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1022                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1023                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1024                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1025                                      [0-9Xx]                 # check digit
1026                                    \b)
1027                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1028                 wfProfileOut( __METHOD__ );
1029                 return $text;
1030         }
1031
1032         function magicLinkCallback( $m ) {
1033                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1034                         # Skip HTML element
1035                         return $m[0];
1036                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1037                         $isbn = $m[2];
1038                         $num = strtr( $isbn, array(
1039                                 '-' => '',
1040                                 ' ' => '',
1041                                 'x' => 'X',
1042                         ));
1043                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1044                         $text = '<a href="' .
1045                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1046                                 "\" class=\"internal\">ISBN $isbn</a>";
1047                 } else {
1048                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1049                                 $keyword = 'RFC';
1050                                 $urlmsg = 'rfcurl';
1051                                 $id = $m[1];
1052                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1053                                 $keyword = 'PMID';
1054                                 $urlmsg = 'pubmedurl';
1055                                 $id = $m[1];
1056                         } else {
1057                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1058                                         substr($m[0], 0, 20 ) . '"' );
1059                         }
1060
1061                         $url = wfMsg( $urlmsg, $id);
1062                         $sk = $this->mOptions->getSkin();
1063                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1064                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1065                 }
1066                 return $text;
1067         }
1068
1069         /**
1070          * Parse headers and return html
1071          *
1072          * @private
1073          */
1074         function doHeadings( $text ) {
1075                 $fname = 'Parser::doHeadings';
1076                 wfProfileIn( $fname );
1077                 for ( $i = 6; $i >= 1; --$i ) {
1078                         $h = str_repeat( '=', $i );
1079                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1080                           "<h{$i}>\\1</h{$i}>\\2", $text );
1081                 }
1082                 wfProfileOut( $fname );
1083                 return $text;
1084         }
1085
1086         /**
1087          * Replace single quotes with HTML markup
1088          * @private
1089          * @return string the altered text
1090          */
1091         function doAllQuotes( $text ) {
1092                 $fname = 'Parser::doAllQuotes';
1093                 wfProfileIn( $fname );
1094                 $outtext = '';
1095                 $lines = explode( "\n", $text );
1096                 foreach ( $lines as $line ) {
1097                         $outtext .= $this->doQuotes ( $line ) . "\n";
1098                 }
1099                 $outtext = substr($outtext, 0,-1);
1100                 wfProfileOut( $fname );
1101                 return $outtext;
1102         }
1103
1104         /**
1105          * Helper function for doAllQuotes()
1106          */
1107         public function doQuotes( $text ) {
1108                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1109                 if ( count( $arr ) == 1 )
1110                         return $text;
1111                 else
1112                 {
1113                         # First, do some preliminary work. This may shift some apostrophes from
1114                         # being mark-up to being text. It also counts the number of occurrences
1115                         # of bold and italics mark-ups.
1116                         $i = 0;
1117                         $numbold = 0;
1118                         $numitalics = 0;
1119                         foreach ( $arr as $r )
1120                         {
1121                                 if ( ( $i % 2 ) == 1 )
1122                                 {
1123                                         # If there are ever four apostrophes, assume the first is supposed to
1124                                         # be text, and the remaining three constitute mark-up for bold text.
1125                                         if ( strlen( $arr[$i] ) == 4 )
1126                                         {
1127                                                 $arr[$i-1] .= "'";
1128                                                 $arr[$i] = "'''";
1129                                         }
1130                                         # If there are more than 5 apostrophes in a row, assume they're all
1131                                         # text except for the last 5.
1132                                         else if ( strlen( $arr[$i] ) > 5 )
1133                                         {
1134                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1135                                                 $arr[$i] = "'''''";
1136                                         }
1137                                         # Count the number of occurrences of bold and italics mark-ups.
1138                                         # We are not counting sequences of five apostrophes.
1139                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1140                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1141                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1142                                 }
1143                                 $i++;
1144                         }
1145
1146                         # If there is an odd number of both bold and italics, it is likely
1147                         # that one of the bold ones was meant to be an apostrophe followed
1148                         # by italics. Which one we cannot know for certain, but it is more
1149                         # likely to be one that has a single-letter word before it.
1150                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1151                         {
1152                                 $i = 0;
1153                                 $firstsingleletterword = -1;
1154                                 $firstmultiletterword = -1;
1155                                 $firstspace = -1;
1156                                 foreach ( $arr as $r )
1157                                 {
1158                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1159                                         {
1160                                                 $x1 = substr ($arr[$i-1], -1);
1161                                                 $x2 = substr ($arr[$i-1], -2, 1);
1162                                                 if ($x1 == ' ') {
1163                                                         if ($firstspace == -1) $firstspace = $i;
1164                                                 } else if ($x2 == ' ') {
1165                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1166                                                 } else {
1167                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1168                                                 }
1169                                         }
1170                                         $i++;
1171                                 }
1172
1173                                 # If there is a single-letter word, use it!
1174                                 if ($firstsingleletterword > -1)
1175                                 {
1176                                         $arr [ $firstsingleletterword ] = "''";
1177                                         $arr [ $firstsingleletterword-1 ] .= "'";
1178                                 }
1179                                 # If not, but there's a multi-letter word, use that one.
1180                                 else if ($firstmultiletterword > -1)
1181                                 {
1182                                         $arr [ $firstmultiletterword ] = "''";
1183                                         $arr [ $firstmultiletterword-1 ] .= "'";
1184                                 }
1185                                 # ... otherwise use the first one that has neither.
1186                                 # (notice that it is possible for all three to be -1 if, for example,
1187                                 # there is only one pentuple-apostrophe in the line)
1188                                 else if ($firstspace > -1)
1189                                 {
1190                                         $arr [ $firstspace ] = "''";
1191                                         $arr [ $firstspace-1 ] .= "'";
1192                                 }
1193                         }
1194
1195                         # Now let's actually convert our apostrophic mush to HTML!
1196                         $output = '';
1197                         $buffer = '';
1198                         $state = '';
1199                         $i = 0;
1200                         foreach ($arr as $r)
1201                         {
1202                                 if (($i % 2) == 0)
1203                                 {
1204                                         if ($state == 'both')
1205                                                 $buffer .= $r;
1206                                         else
1207                                                 $output .= $r;
1208                                 }
1209                                 else
1210                                 {
1211                                         if (strlen ($r) == 2)
1212                                         {
1213                                                 if ($state == 'i')
1214                                                 { $output .= '</i>'; $state = ''; }
1215                                                 else if ($state == 'bi')
1216                                                 { $output .= '</i>'; $state = 'b'; }
1217                                                 else if ($state == 'ib')
1218                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1219                                                 else if ($state == 'both')
1220                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1221                                                 else # $state can be 'b' or ''
1222                                                 { $output .= '<i>'; $state .= 'i'; }
1223                                         }
1224                                         else if (strlen ($r) == 3)
1225                                         {
1226                                                 if ($state == 'b')
1227                                                 { $output .= '</b>'; $state = ''; }
1228                                                 else if ($state == 'bi')
1229                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1230                                                 else if ($state == 'ib')
1231                                                 { $output .= '</b>'; $state = 'i'; }
1232                                                 else if ($state == 'both')
1233                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1234                                                 else # $state can be 'i' or ''
1235                                                 { $output .= '<b>'; $state .= 'b'; }
1236                                         }
1237                                         else if (strlen ($r) == 5)
1238                                         {
1239                                                 if ($state == 'b')
1240                                                 { $output .= '</b><i>'; $state = 'i'; }
1241                                                 else if ($state == 'i')
1242                                                 { $output .= '</i><b>'; $state = 'b'; }
1243                                                 else if ($state == 'bi')
1244                                                 { $output .= '</i></b>'; $state = ''; }
1245                                                 else if ($state == 'ib')
1246                                                 { $output .= '</b></i>'; $state = ''; }
1247                                                 else if ($state == 'both')
1248                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1249                                                 else # ($state == '')
1250                                                 { $buffer = ''; $state = 'both'; }
1251                                         }
1252                                 }
1253                                 $i++;
1254                         }
1255                         # Now close all remaining tags.  Notice that the order is important.
1256                         if ($state == 'b' || $state == 'ib')
1257                                 $output .= '</b>';
1258                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1259                                 $output .= '</i>';
1260                         if ($state == 'bi')
1261                                 $output .= '</b>';
1262                         # There might be lonely ''''', so make sure we have a buffer
1263                         if ($state == 'both' && $buffer)
1264                                 $output .= '<b><i>'.$buffer.'</i></b>';
1265                         return $output;
1266                 }
1267         }
1268
1269         /**
1270          * Replace external links
1271          *
1272          * Note: this is all very hackish and the order of execution matters a lot.
1273          * Make sure to run maintenance/parserTests.php if you change this code.
1274          *
1275          * @private
1276          */
1277         function replaceExternalLinks( $text ) {
1278                 global $wgContLang;
1279                 $fname = 'Parser::replaceExternalLinks';
1280                 wfProfileIn( $fname );
1281
1282                 $sk = $this->mOptions->getSkin();
1283
1284                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1285
1286                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1287
1288                 $i = 0;
1289                 while ( $i<count( $bits ) ) {
1290                         $url = $bits[$i++];
1291                         $protocol = $bits[$i++];
1292                         $text = $bits[$i++];
1293                         $trail = $bits[$i++];
1294
1295                         # The characters '<' and '>' (which were escaped by
1296                         # removeHTMLtags()) should not be included in
1297                         # URLs, per RFC 2396.
1298                         $m2 = array();
1299                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1300                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1301                                 $url = substr($url, 0, $m2[0][1]);
1302                         }
1303
1304                         # If the link text is an image URL, replace it with an <img> tag
1305                         # This happened by accident in the original parser, but some people used it extensively
1306                         $img = $this->maybeMakeExternalImage( $text );
1307                         if ( $img !== false ) {
1308                                 $text = $img;
1309                         }
1310
1311                         $dtrail = '';
1312
1313                         # Set linktype for CSS - if URL==text, link is essentially free
1314                         $linktype = ($text == $url) ? 'free' : 'text';
1315
1316                         # No link text, e.g. [http://domain.tld/some.link]
1317                         if ( $text == '' ) {
1318                                 # Autonumber if allowed. See bug #5918
1319                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1320                                         $text = '[' . ++$this->mAutonumber . ']';
1321                                         $linktype = 'autonumber';
1322                                 } else {
1323                                         # Otherwise just use the URL
1324                                         $text = htmlspecialchars( $url );
1325                                         $linktype = 'free';
1326                                 }
1327                         } else {
1328                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1329                                 # Check for trail
1330                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1331                         }
1332
1333                         $text = $wgContLang->markNoConversion($text);
1334
1335                         $url = Sanitizer::cleanUrl( $url );
1336
1337                         # Process the trail (i.e. everything after this link up until start of the next link),
1338                         # replacing any non-bracketed links
1339                         $trail = $this->replaceFreeExternalLinks( $trail );
1340
1341                         # Use the encoded URL
1342                         # This means that users can paste URLs directly into the text
1343                         # Funny characters like &ouml; aren't valid in URLs anyway
1344                         # This was changed in August 2004
1345                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1346
1347                         # Register link in the output object.
1348                         # Replace unnecessary URL escape codes with the referenced character
1349                         # This prevents spammers from hiding links from the filters
1350                         $pasteurized = self::replaceUnusualEscapes( $url );
1351                         $this->mOutput->addExternalLink( $pasteurized );
1352                 }
1353
1354                 wfProfileOut( $fname );
1355                 return $s;
1356         }
1357
1358         /**
1359          * Replace anything that looks like a URL with a link
1360          * @private
1361          */
1362         function replaceFreeExternalLinks( $text ) {
1363                 global $wgContLang;
1364                 $fname = 'Parser::replaceFreeExternalLinks';
1365                 wfProfileIn( $fname );
1366
1367                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1368                 $s = array_shift( $bits );
1369                 $i = 0;
1370
1371                 $sk = $this->mOptions->getSkin();
1372
1373                 while ( $i < count( $bits ) ){
1374                         $protocol = $bits[$i++];
1375                         $remainder = $bits[$i++];
1376
1377                         $m = array();
1378                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1379                                 # Found some characters after the protocol that look promising
1380                                 $url = $protocol . $m[1];
1381                                 $trail = $m[2];
1382
1383                                 # special case: handle urls as url args:
1384                                 # http://www.example.com/foo?=http://www.example.com/bar
1385                                 if(strlen($trail) == 0 &&
1386                                         isset($bits[$i]) &&
1387                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1388                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1389                                 {
1390                                         # add protocol, arg
1391                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1392                                         $i += 2;
1393                                         $trail = $m[2];
1394                                 }
1395
1396                                 # The characters '<' and '>' (which were escaped by
1397                                 # removeHTMLtags()) should not be included in
1398                                 # URLs, per RFC 2396.
1399                                 $m2 = array();
1400                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1401                                         $trail = substr($url, $m2[0][1]) . $trail;
1402                                         $url = substr($url, 0, $m2[0][1]);
1403                                 }
1404
1405                                 # Move trailing punctuation to $trail
1406                                 $sep = ',;\.:!?';
1407                                 # If there is no left bracket, then consider right brackets fair game too
1408                                 if ( strpos( $url, '(' ) === false ) {
1409                                         $sep .= ')';
1410                                 }
1411
1412                                 $numSepChars = strspn( strrev( $url ), $sep );
1413                                 if ( $numSepChars ) {
1414                                         $trail = substr( $url, -$numSepChars ) . $trail;
1415                                         $url = substr( $url, 0, -$numSepChars );
1416                                 }
1417
1418                                 $url = Sanitizer::cleanUrl( $url );
1419
1420                                 # Is this an external image?
1421                                 $text = $this->maybeMakeExternalImage( $url );
1422                                 if ( $text === false ) {
1423                                         # Not an image, make a link
1424                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1425                                         # Register it in the output object...
1426                                         # Replace unnecessary URL escape codes with their equivalent characters
1427                                         $pasteurized = self::replaceUnusualEscapes( $url );
1428                                         $this->mOutput->addExternalLink( $pasteurized );
1429                                 }
1430                                 $s .= $text . $trail;
1431                         } else {
1432                                 $s .= $protocol . $remainder;
1433                         }
1434                 }
1435                 wfProfileOut( $fname );
1436                 return $s;
1437         }
1438
1439         /**
1440          * Replace unusual URL escape codes with their equivalent characters
1441          * @param string
1442          * @return string
1443          * @static
1444          * @todo  This can merge genuinely required bits in the path or query string,
1445          *        breaking legit URLs. A proper fix would treat the various parts of
1446          *        the URL differently; as a workaround, just use the output for
1447          *        statistical records, not for actual linking/output.
1448          */
1449         static function replaceUnusualEscapes( $url ) {
1450                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1451                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1452         }
1453
1454         /**
1455          * Callback function used in replaceUnusualEscapes().
1456          * Replaces unusual URL escape codes with their equivalent character
1457          * @static
1458          * @private
1459          */
1460         private static function replaceUnusualEscapesCallback( $matches ) {
1461                 $char = urldecode( $matches[0] );
1462                 $ord = ord( $char );
1463                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1464                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1465                         // No, shouldn't be escaped
1466                         return $char;
1467                 } else {
1468                         // Yes, leave it escaped
1469                         return $matches[0];
1470                 }
1471         }
1472
1473         /**
1474          * make an image if it's allowed, either through the global
1475          * option or through the exception
1476          * @private
1477          */
1478         function maybeMakeExternalImage( $url ) {
1479                 $sk = $this->mOptions->getSkin();
1480                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1481                 $imagesexception = !empty($imagesfrom);
1482                 $text = false;
1483                 if ( $this->mOptions->getAllowExternalImages()
1484                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1485                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1486                                 # Image found
1487                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1488                         }
1489                 }
1490                 return $text;
1491         }
1492
1493         /**
1494          * Process [[ ]] wikilinks
1495          *
1496          * @private
1497          */
1498         function replaceInternalLinks( $s ) {
1499                 global $wgContLang;
1500                 static $fname = 'Parser::replaceInternalLinks' ;
1501
1502                 wfProfileIn( $fname );
1503
1504                 wfProfileIn( $fname.'-setup' );
1505                 static $tc = FALSE;
1506                 # the % is needed to support urlencoded titles as well
1507                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1508
1509                 $sk = $this->mOptions->getSkin();
1510
1511                 #split the entire text string on occurences of [[
1512                 $a = explode( '[[', ' ' . $s );
1513                 #get the first element (all text up to first [[), and remove the space we added
1514                 $s = array_shift( $a );
1515                 $s = substr( $s, 1 );
1516
1517                 # Match a link having the form [[namespace:link|alternate]]trail
1518                 static $e1 = FALSE;
1519                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1520                 # Match cases where there is no "]]", which might still be images
1521                 static $e1_img = FALSE;
1522                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1523                 # Match the end of a line for a word that's not followed by whitespace,
1524                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1525                 $e2 = wfMsgForContent( 'linkprefix' );
1526
1527                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1528                 if( is_null( $this->mTitle ) ) {
1529                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1530                 }
1531                 $nottalk = !$this->mTitle->isTalkPage();
1532
1533                 if ( $useLinkPrefixExtension ) {
1534                         $m = array();
1535                         if ( preg_match( $e2, $s, $m ) ) {
1536                                 $first_prefix = $m[2];
1537                         } else {
1538                                 $first_prefix = false;
1539                         }
1540                 } else {
1541                         $prefix = '';
1542                 }
1543
1544                 if($wgContLang->hasVariants()) {
1545                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1546                 } else {
1547                         $selflink = array($this->mTitle->getPrefixedText());
1548                 }
1549                 $useSubpages = $this->areSubpagesAllowed();
1550                 wfProfileOut( $fname.'-setup' );
1551
1552                 # Loop for each link
1553                 for ($k = 0; isset( $a[$k] ); $k++) {
1554                         $line = $a[$k];
1555                         if ( $useLinkPrefixExtension ) {
1556                                 wfProfileIn( $fname.'-prefixhandling' );
1557                                 if ( preg_match( $e2, $s, $m ) ) {
1558                                         $prefix = $m[2];
1559                                         $s = $m[1];
1560                                 } else {
1561                                         $prefix='';
1562                                 }
1563                                 # first link
1564                                 if($first_prefix) {
1565                                         $prefix = $first_prefix;
1566                                         $first_prefix = false;
1567                                 }
1568                                 wfProfileOut( $fname.'-prefixhandling' );
1569                         }
1570
1571                         $might_be_img = false;
1572
1573                         wfProfileIn( "$fname-e1" );
1574                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1575                                 $text = $m[2];
1576                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1577                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1578                                 # the real problem is with the $e1 regex
1579                                 # See bug 1300.
1580                                 #
1581                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1582                                 # and no image is in sight. See bug 2095.
1583                                 #
1584                                 if( $text !== '' &&
1585                                         substr( $m[3], 0, 1 ) === ']' &&
1586                                         strpos($text, '[') !== false
1587                                 )
1588                                 {
1589                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1590                                         $m[3] = substr( $m[3], 1 );
1591                                 }
1592                                 # fix up urlencoded title texts
1593                                 if( strpos( $m[1], '%' ) !== false ) {
1594                                         # Should anchors '#' also be rejected?
1595                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1596                                 }
1597                                 $trail = $m[3];
1598                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1599                                 $might_be_img = true;
1600                                 $text = $m[2];
1601                                 if ( strpos( $m[1], '%' ) !== false ) {
1602                                         $m[1] = urldecode($m[1]);
1603                                 }
1604                                 $trail = "";
1605                         } else { # Invalid form; output directly
1606                                 $s .= $prefix . '[[' . $line ;
1607                                 wfProfileOut( "$fname-e1" );
1608                                 continue;
1609                         }
1610                         wfProfileOut( "$fname-e1" );
1611                         wfProfileIn( "$fname-misc" );
1612
1613                         # Don't allow internal links to pages containing
1614                         # PROTO: where PROTO is a valid URL protocol; these
1615                         # should be external links.
1616                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1617                                 $s .= $prefix . '[[' . $line ;
1618                                 continue;
1619                         }
1620
1621                         # Make subpage if necessary
1622                         if( $useSubpages ) {
1623                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1624                         } else {
1625                                 $link = $m[1];
1626                         }
1627
1628                         $noforce = (substr($m[1], 0, 1) != ':');
1629                         if (!$noforce) {
1630                                 # Strip off leading ':'
1631                                 $link = substr($link, 1);
1632                         }
1633
1634                         wfProfileOut( "$fname-misc" );
1635                         wfProfileIn( "$fname-title" );
1636                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1637                         if( !$nt ) {
1638                                 $s .= $prefix . '[[' . $line;
1639                                 wfProfileOut( "$fname-title" );
1640                                 continue;
1641                         }
1642
1643                         $ns = $nt->getNamespace();
1644                         $iw = $nt->getInterWiki();
1645                         wfProfileOut( "$fname-title" );
1646
1647                         if ($might_be_img) { # if this is actually an invalid link
1648                                 wfProfileIn( "$fname-might_be_img" );
1649                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1650                                         $found = false;
1651                                         while (isset ($a[$k+1]) ) {
1652                                                 #look at the next 'line' to see if we can close it there
1653                                                 $spliced = array_splice( $a, $k + 1, 1 );
1654                                                 $next_line = array_shift( $spliced );
1655                                                 $m = explode( ']]', $next_line, 3 );
1656                                                 if ( count( $m ) == 3 ) {
1657                                                         # the first ]] closes the inner link, the second the image
1658                                                         $found = true;
1659                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1660                                                         $trail = $m[2];
1661                                                         break;
1662                                                 } elseif ( count( $m ) == 2 ) {
1663                                                         #if there's exactly one ]] that's fine, we'll keep looking
1664                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1665                                                 } else {
1666                                                         #if $next_line is invalid too, we need look no further
1667                                                         $text .= '[[' . $next_line;
1668                                                         break;
1669                                                 }
1670                                         }
1671                                         if ( !$found ) {
1672                                                 # we couldn't find the end of this imageLink, so output it raw
1673                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1674                                                 $text = $this->replaceInternalLinks($text);
1675                                                 $s .= "{$prefix}[[$link|$text";
1676                                                 # note: no $trail, because without an end, there *is* no trail
1677                                                 wfProfileOut( "$fname-might_be_img" );
1678                                                 continue;
1679                                         }
1680                                 } else { #it's not an image, so output it raw
1681                                         $s .= "{$prefix}[[$link|$text";
1682                                         # note: no $trail, because without an end, there *is* no trail
1683                                         wfProfileOut( "$fname-might_be_img" );
1684                                         continue;
1685                                 }
1686                                 wfProfileOut( "$fname-might_be_img" );
1687                         }
1688
1689                         $wasblank = ( '' == $text );
1690                         if( $wasblank ) $text = $link;
1691
1692                         # Link not escaped by : , create the various objects
1693                         if( $noforce ) {
1694
1695                                 # Interwikis
1696                                 wfProfileIn( "$fname-interwiki" );
1697                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1698                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1699                                         $s = rtrim($s . $prefix);
1700                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1701                                         wfProfileOut( "$fname-interwiki" );
1702                                         continue;
1703                                 }
1704                                 wfProfileOut( "$fname-interwiki" );
1705
1706                                 if ( $ns == NS_IMAGE ) {
1707                                         wfProfileIn( "$fname-image" );
1708                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1709                                                 # recursively parse links inside the image caption
1710                                                 # actually, this will parse them in any other parameters, too,
1711                                                 # but it might be hard to fix that, and it doesn't matter ATM
1712                                                 $text = $this->replaceExternalLinks($text);
1713                                                 $text = $this->replaceInternalLinks($text);
1714
1715                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1716                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1717                                                 $this->mOutput->addImage( $nt->getDBkey() );
1718
1719                                                 wfProfileOut( "$fname-image" );
1720                                                 continue;
1721                                         } else {
1722                                                 # We still need to record the image's presence on the page
1723                                                 $this->mOutput->addImage( $nt->getDBkey() );
1724                                         }
1725                                         wfProfileOut( "$fname-image" );
1726
1727                                 }
1728
1729                                 if ( $ns == NS_CATEGORY ) {
1730                                         wfProfileIn( "$fname-category" );
1731                                         $s = rtrim($s . "\n"); # bug 87
1732
1733                                         if ( $wasblank ) {
1734                                                 $sortkey = $this->getDefaultSort();
1735                                         } else {
1736                                                 $sortkey = $text;
1737                                         }
1738                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1739                                         $sortkey = str_replace( "\n", '', $sortkey );
1740                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1741                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1742
1743                                         /**
1744                                          * Strip the whitespace Category links produce, see bug 87
1745                                          * @todo We might want to use trim($tmp, "\n") here.
1746                                          */
1747                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1748
1749                                         wfProfileOut( "$fname-category" );
1750                                         continue;
1751                                 }
1752                         }
1753
1754                         # Self-link checking
1755                         if( $nt->getFragment() === '' ) {
1756                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1757                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1758                                         continue;
1759                                 }
1760                         }
1761
1762                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1763                         if( $ns == NS_MEDIA ) {
1764                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1765                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1766                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1767                                 $this->mOutput->addImage( $nt->getDBkey() );
1768                                 continue;
1769                         } elseif( $ns == NS_SPECIAL ) {
1770                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1771                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1772                                 } else {
1773                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1774                                 }
1775                                 continue;
1776                         } elseif( $ns == NS_IMAGE ) {
1777                                 $img = wfFindFile( $nt );
1778                                 if( $img ) {
1779                                         // Force a blue link if the file exists; may be a remote
1780                                         // upload on the shared repository, and we want to see its
1781                                         // auto-generated page.
1782                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1783                                         $this->mOutput->addLink( $nt );
1784                                         continue;
1785                                 }
1786                         }
1787                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1788                 }
1789                 wfProfileOut( $fname );
1790                 return $s;
1791         }
1792
1793         /**
1794          * Make a link placeholder. The text returned can be later resolved to a real link with
1795          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1796          * parsing of interwiki links, and secondly to allow all existence checks and
1797          * article length checks (for stub links) to be bundled into a single query.
1798          *
1799          */
1800         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1801                 wfProfileIn( __METHOD__ );
1802                 if ( ! is_object($nt) ) {
1803                         # Fail gracefully
1804                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1805                 } else {
1806                         # Separate the link trail from the rest of the link
1807                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1808
1809                         if ( $nt->isExternal() ) {
1810                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1811                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1812                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1813                         } else {
1814                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1815                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1816                                 $this->mLinkHolders['queries'][] = $query;
1817                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1818                                 $this->mLinkHolders['titles'][] = $nt;
1819
1820                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1821                         }
1822                 }
1823                 wfProfileOut( __METHOD__ );
1824                 return $retVal;
1825         }
1826
1827         /**
1828          * Render a forced-blue link inline; protect against double expansion of
1829          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1830          * Since this little disaster has to split off the trail text to avoid
1831          * breaking URLs in the following text without breaking trails on the
1832          * wiki links, it's been made into a horrible function.
1833          *
1834          * @param Title $nt
1835          * @param string $text
1836          * @param string $query
1837          * @param string $trail
1838          * @param string $prefix
1839          * @return string HTML-wikitext mix oh yuck
1840          */
1841         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1842                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1843                 $sk = $this->mOptions->getSkin();
1844                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1845                 return $this->armorLinks( $link ) . $trail;
1846         }
1847
1848         /**
1849          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1850          * going to go through further parsing steps before inline URL expansion.
1851          *
1852          * In particular this is important when using action=render, which causes
1853          * full URLs to be included.
1854          *
1855          * Oh man I hate our multi-layer parser!
1856          *
1857          * @param string more-or-less HTML
1858          * @return string less-or-more HTML with NOPARSE bits
1859          */
1860         function armorLinks( $text ) {
1861                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1862                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1863         }
1864
1865         /**
1866          * Return true if subpage links should be expanded on this page.
1867          * @return bool
1868          */
1869         function areSubpagesAllowed() {
1870                 # Some namespaces don't allow subpages
1871                 global $wgNamespacesWithSubpages;
1872                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1873         }
1874
1875         /**
1876          * Handle link to subpage if necessary
1877          * @param string $target the source of the link
1878          * @param string &$text the link text, modified as necessary
1879          * @return string the full name of the link
1880          * @private
1881          */
1882         function maybeDoSubpageLink($target, &$text) {
1883                 # Valid link forms:
1884                 # Foobar -- normal
1885                 # :Foobar -- override special treatment of prefix (images, language links)
1886                 # /Foobar -- convert to CurrentPage/Foobar
1887                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1888                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1889                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1890
1891                 $fname = 'Parser::maybeDoSubpageLink';
1892                 wfProfileIn( $fname );
1893                 $ret = $target; # default return value is no change
1894
1895                 # Some namespaces don't allow subpages,
1896                 # so only perform processing if subpages are allowed
1897                 if( $this->areSubpagesAllowed() ) {
1898                         $hash = strpos( $target, '#' );
1899                         if( $hash !== false ) {
1900                                 $suffix = substr( $target, $hash );
1901                                 $target = substr( $target, 0, $hash );
1902                         } else {
1903                                 $suffix = '';
1904                         }
1905                         # bug 7425
1906                         $target = trim( $target );
1907                         # Look at the first character
1908                         if( $target != '' && $target{0} == '/' ) {
1909                                 # / at end means we don't want the slash to be shown
1910                                 $m = array();
1911                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1912                                 if( $trailingSlashes ) {
1913                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1914                                 } else {
1915                                         $noslash = substr( $target, 1 );
1916                                 }
1917
1918                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1919                                 if( '' === $text ) {
1920                                         $text = $target . $suffix;
1921                                 } # this might be changed for ugliness reasons
1922                         } else {
1923                                 # check for .. subpage backlinks
1924                                 $dotdotcount = 0;
1925                                 $nodotdot = $target;
1926                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1927                                         ++$dotdotcount;
1928                                         $nodotdot = substr( $nodotdot, 3 );
1929                                 }
1930                                 if($dotdotcount > 0) {
1931                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1932                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1933                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1934                                                 # / at the end means don't show full path
1935                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1936                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1937                                                         if( '' === $text ) {
1938                                                                 $text = $nodotdot . $suffix;
1939                                                         }
1940                                                 }
1941                                                 $nodotdot = trim( $nodotdot );
1942                                                 if( $nodotdot != '' ) {
1943                                                         $ret .= '/' . $nodotdot;
1944                                                 }
1945                                                 $ret .= $suffix;
1946                                         }
1947                                 }
1948                         }
1949                 }
1950
1951                 wfProfileOut( $fname );
1952                 return $ret;
1953         }
1954
1955         /**#@+
1956          * Used by doBlockLevels()
1957          * @private
1958          */
1959         /* private */ function closeParagraph() {
1960                 $result = '';
1961                 if ( '' != $this->mLastSection ) {
1962                         $result = '</' . $this->mLastSection  . ">\n";
1963                 }
1964                 $this->mInPre = false;
1965                 $this->mLastSection = '';
1966                 return $result;
1967         }
1968         # getCommon() returns the length of the longest common substring
1969         # of both arguments, starting at the beginning of both.
1970         #
1971         /* private */ function getCommon( $st1, $st2 ) {
1972                 $fl = strlen( $st1 );
1973                 $shorter = strlen( $st2 );
1974                 if ( $fl < $shorter ) { $shorter = $fl; }
1975
1976                 for ( $i = 0; $i < $shorter; ++$i ) {
1977                         if ( $st1{$i} != $st2{$i} ) { break; }
1978                 }
1979                 return $i;
1980         }
1981         # These next three functions open, continue, and close the list
1982         # element appropriate to the prefix character passed into them.
1983         #
1984         /* private */ function openList( $char ) {
1985                 $result = $this->closeParagraph();
1986
1987                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1988                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1989                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1990                 else if ( ';' == $char ) {
1991                         $result .= '<dl><dt>';
1992                         $this->mDTopen = true;
1993                 }
1994                 else { $result = '<!-- ERR 1 -->'; }
1995
1996                 return $result;
1997         }
1998
1999         /* private */ function nextItem( $char ) {
2000                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2001                 else if ( ':' == $char || ';' == $char ) {
2002                         $close = '</dd>';
2003                         if ( $this->mDTopen ) { $close = '</dt>'; }
2004                         if ( ';' == $char ) {
2005                                 $this->mDTopen = true;
2006                                 return $close . '<dt>';
2007                         } else {
2008                                 $this->mDTopen = false;
2009                                 return $close . '<dd>';
2010                         }
2011                 }
2012                 return '<!-- ERR 2 -->';
2013         }
2014
2015         /* private */ function closeList( $char ) {
2016                 if ( '*' == $char ) { $text = '</li></ul>'; }
2017                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2018                 else if ( ':' == $char ) {
2019                         if ( $this->mDTopen ) {
2020                                 $this->mDTopen = false;
2021                                 $text = '</dt></dl>';
2022                         } else {
2023                                 $text = '</dd></dl>';
2024                         }
2025                 }
2026                 else {  return '<!-- ERR 3 -->'; }
2027                 return $text."\n";
2028         }
2029         /**#@-*/
2030
2031         /**
2032          * Make lists from lines starting with ':', '*', '#', etc.
2033          *
2034          * @private
2035          * @return string the lists rendered as HTML
2036          */
2037         function doBlockLevels( $text, $linestart ) {
2038                 $fname = 'Parser::doBlockLevels';
2039                 wfProfileIn( $fname );
2040
2041                 # Parsing through the text line by line.  The main thing
2042                 # happening here is handling of block-level elements p, pre,
2043                 # and making lists from lines starting with * # : etc.
2044                 #
2045                 $textLines = explode( "\n", $text );
2046
2047                 $lastPrefix = $output = '';
2048                 $this->mDTopen = $inBlockElem = false;
2049                 $prefixLength = 0;
2050                 $paragraphStack = false;
2051
2052                 if ( !$linestart ) {
2053                         $output .= array_shift( $textLines );
2054                 }
2055                 foreach ( $textLines as $oLine ) {
2056                         $lastPrefixLength = strlen( $lastPrefix );
2057                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2058                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2059                         if ( !$this->mInPre ) {
2060                                 # Multiple prefixes may abut each other for nested lists.
2061                                 $prefixLength = strspn( $oLine, '*#:;' );
2062                                 $pref = substr( $oLine, 0, $prefixLength );
2063
2064                                 # eh?
2065                                 $pref2 = str_replace( ';', ':', $pref );
2066                                 $t = substr( $oLine, $prefixLength );
2067                                 $this->mInPre = !empty($preOpenMatch);
2068                         } else {
2069                                 # Don't interpret any other prefixes in preformatted text
2070                                 $prefixLength = 0;
2071                                 $pref = $pref2 = '';
2072                                 $t = $oLine;
2073                         }
2074
2075                         # List generation
2076                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2077                                 # Same as the last item, so no need to deal with nesting or opening stuff
2078                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2079                                 $paragraphStack = false;
2080
2081                                 if ( substr( $pref, -1 ) == ';') {
2082                                         # The one nasty exception: definition lists work like this:
2083                                         # ; title : definition text
2084                                         # So we check for : in the remainder text to split up the
2085                                         # title and definition, without b0rking links.
2086                                         $term = $t2 = '';
2087                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2088                                                 $t = $t2;
2089                                                 $output .= $term . $this->nextItem( ':' );
2090                                         }
2091                                 }
2092                         } elseif( $prefixLength || $lastPrefixLength ) {
2093                                 # Either open or close a level...
2094                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2095                                 $paragraphStack = false;
2096
2097                                 while( $commonPrefixLength < $lastPrefixLength ) {
2098                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2099                                         --$lastPrefixLength;
2100                                 }
2101                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2102                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2103                                 }
2104                                 while ( $prefixLength > $commonPrefixLength ) {
2105                                         $char = substr( $pref, $commonPrefixLength, 1 );
2106                                         $output .= $this->openList( $char );
2107
2108                                         if ( ';' == $char ) {
2109                                                 # FIXME: This is dupe of code above
2110                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2111                                                         $t = $t2;
2112                                                         $output .= $term . $this->nextItem( ':' );
2113                                                 }
2114                                         }
2115                                         ++$commonPrefixLength;
2116                                 }
2117                                 $lastPrefix = $pref2;
2118                         }
2119                         if( 0 == $prefixLength ) {
2120                                 wfProfileIn( "$fname-paragraph" );
2121                                 # No prefix (not in list)--go to paragraph mode
2122                                 // XXX: use a stack for nestable elements like span, table and div
2123                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2124                                 $closematch = preg_match(
2125                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2126                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2127                                 if ( $openmatch or $closematch ) {
2128                                         $paragraphStack = false;
2129                                         # TODO bug 5718: paragraph closed
2130                                         $output .= $this->closeParagraph();
2131                                         if ( $preOpenMatch and !$preCloseMatch ) {
2132                                                 $this->mInPre = true;
2133                                         }
2134                                         if ( $closematch ) {
2135                                                 $inBlockElem = false;
2136                                         } else {
2137                                                 $inBlockElem = true;
2138                                         }
2139                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2140                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2141                                                 // pre
2142                                                 if ($this->mLastSection != 'pre') {
2143                                                         $paragraphStack = false;
2144                                                         $output .= $this->closeParagraph().'<pre>';
2145                                                         $this->mLastSection = 'pre';
2146                                                 }
2147                                                 $t = substr( $t, 1 );
2148                                         } else {
2149                                                 // paragraph
2150                                                 if ( '' == trim($t) ) {
2151                                                         if ( $paragraphStack ) {
2152                                                                 $output .= $paragraphStack.'<br />';
2153                                                                 $paragraphStack = false;
2154                                                                 $this->mLastSection = 'p';
2155                                                         } else {
2156                                                                 if ($this->mLastSection != 'p' ) {
2157                                                                         $output .= $this->closeParagraph();
2158                                                                         $this->mLastSection = '';
2159                                                                         $paragraphStack = '<p>';
2160                                                                 } else {
2161                                                                         $paragraphStack = '</p><p>';
2162                                                                 }
2163                                                         }
2164                                                 } else {
2165                                                         if ( $paragraphStack ) {
2166                                                                 $output .= $paragraphStack;
2167                                                                 $paragraphStack = false;
2168                                                                 $this->mLastSection = 'p';
2169                                                         } else if ($this->mLastSection != 'p') {
2170                                                                 $output .= $this->closeParagraph().'<p>';
2171                                                                 $this->mLastSection = 'p';
2172                                                         }
2173                                                 }
2174                                         }
2175                                 }
2176                                 wfProfileOut( "$fname-paragraph" );
2177                         }
2178                         // somewhere above we forget to get out of pre block (bug 785)
2179                         if($preCloseMatch && $this->mInPre) {
2180                                 $this->mInPre = false;
2181                         }
2182                         if ($paragraphStack === false) {
2183                                 $output .= $t."\n";
2184                         }
2185                 }
2186                 while ( $prefixLength ) {
2187                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2188                         --$prefixLength;
2189                 }
2190                 if ( '' != $this->mLastSection ) {
2191                         $output .= '</' . $this->mLastSection . '>';
2192                         $this->mLastSection = '';
2193                 }
2194
2195                 wfProfileOut( $fname );
2196                 return $output;
2197         }
2198
2199         /**
2200          * Split up a string on ':', ignoring any occurences inside tags
2201          * to prevent illegal overlapping.
2202          * @param string $str the string to split
2203          * @param string &$before set to everything before the ':'
2204          * @param string &$after set to everything after the ':'
2205          * return string the position of the ':', or false if none found
2206          */
2207         function findColonNoLinks($str, &$before, &$after) {
2208                 $fname = 'Parser::findColonNoLinks';
2209                 wfProfileIn( $fname );
2210
2211                 $pos = strpos( $str, ':' );
2212                 if( $pos === false ) {
2213                         // Nothing to find!
2214                         wfProfileOut( $fname );
2215                         return false;
2216                 }
2217
2218                 $lt = strpos( $str, '<' );
2219                 if( $lt === false || $lt > $pos ) {
2220                         // Easy; no tag nesting to worry about
2221                         $before = substr( $str, 0, $pos );
2222                         $after = substr( $str, $pos+1 );
2223                         wfProfileOut( $fname );
2224                         return $pos;
2225                 }
2226
2227                 // Ugly state machine to walk through avoiding tags.
2228                 $state = self::COLON_STATE_TEXT;
2229                 $stack = 0;
2230                 $len = strlen( $str );
2231                 for( $i = 0; $i < $len; $i++ ) {
2232                         $c = $str{$i};
2233
2234                         switch( $state ) {
2235                         // (Using the number is a performance hack for common cases)
2236                         case 0: // self::COLON_STATE_TEXT:
2237                                 switch( $c ) {
2238                                 case "<":
2239                                         // Could be either a <start> tag or an </end> tag
2240                                         $state = self::COLON_STATE_TAGSTART;
2241                                         break;
2242                                 case ":":
2243                                         if( $stack == 0 ) {
2244                                                 // We found it!
2245                                                 $before = substr( $str, 0, $i );
2246                                                 $after = substr( $str, $i + 1 );
2247                                                 wfProfileOut( $fname );
2248                                                 return $i;
2249                                         }
2250                                         // Embedded in a tag; don't break it.
2251                                         break;
2252                                 default:
2253                                         // Skip ahead looking for something interesting
2254                                         $colon = strpos( $str, ':', $i );
2255                                         if( $colon === false ) {
2256                                                 // Nothing else interesting
2257                                                 wfProfileOut( $fname );
2258                                                 return false;
2259                                         }
2260                                         $lt = strpos( $str, '<', $i );
2261                                         if( $stack === 0 ) {
2262                                                 if( $lt === false || $colon < $lt ) {
2263                                                         // We found it!
2264                                                         $before = substr( $str, 0, $colon );
2265                                                         $after = substr( $str, $colon + 1 );
2266                                                         wfProfileOut( $fname );
2267                                                         return $i;
2268                                                 }
2269                                         }
2270                                         if( $lt === false ) {
2271                                                 // Nothing else interesting to find; abort!
2272                                                 // We're nested, but there's no close tags left. Abort!
2273                                                 break 2;
2274                                         }
2275                                         // Skip ahead to next tag start
2276                                         $i = $lt;
2277                                         $state = self::COLON_STATE_TAGSTART;
2278                                 }
2279                                 break;
2280                         case 1: // self::COLON_STATE_TAG:
2281                                 // In a <tag>
2282                                 switch( $c ) {
2283                                 case ">":
2284                                         $stack++;
2285                                         $state = self::COLON_STATE_TEXT;
2286                                         break;
2287                                 case "/":
2288                                         // Slash may be followed by >?
2289                                         $state = self::COLON_STATE_TAGSLASH;
2290                                         break;
2291                                 default:
2292                                         // ignore
2293                                 }
2294                                 break;
2295                         case 2: // self::COLON_STATE_TAGSTART:
2296                                 switch( $c ) {
2297                                 case "/":
2298                                         $state = self::COLON_STATE_CLOSETAG;
2299                                         break;
2300                                 case "!":
2301                                         $state = self::COLON_STATE_COMMENT;
2302                                         break;
2303                                 case ">":
2304                                         // Illegal early close? This shouldn't happen D:
2305                                         $state = self::COLON_STATE_TEXT;
2306                                         break;
2307                                 default:
2308                                         $state = self::COLON_STATE_TAG;
2309                                 }
2310                                 break;
2311                         case 3: // self::COLON_STATE_CLOSETAG:
2312                                 // In a </tag>
2313                                 if( $c == ">" ) {
2314                                         $stack--;
2315                                         if( $stack < 0 ) {
2316                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2317                                                 wfProfileOut( $fname );
2318                                                 return false;
2319                                         }
2320                                         $state = self::COLON_STATE_TEXT;
2321                                 }
2322                                 break;
2323                         case self::COLON_STATE_TAGSLASH:
2324                                 if( $c == ">" ) {
2325                                         // Yes, a self-closed tag <blah/>
2326                                         $state = self::COLON_STATE_TEXT;
2327                                 } else {
2328                                         // Probably we're jumping the gun, and this is an attribute
2329                                         $state = self::COLON_STATE_TAG;
2330                                 }
2331                                 break;
2332                         case 5: // self::COLON_STATE_COMMENT:
2333                                 if( $c == "-" ) {
2334                                         $state = self::COLON_STATE_COMMENTDASH;
2335                                 }
2336                                 break;
2337                         case self::COLON_STATE_COMMENTDASH:
2338                                 if( $c == "-" ) {
2339                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2340                                 } else {
2341                                         $state = self::COLON_STATE_COMMENT;
2342                                 }
2343                                 break;
2344                         case self::COLON_STATE_COMMENTDASHDASH:
2345                                 if( $c == ">" ) {
2346                                         $state = self::COLON_STATE_TEXT;
2347                                 } else {
2348                                         $state = self::COLON_STATE_COMMENT;
2349                                 }
2350                                 break;
2351                         default:
2352                                 throw new MWException( "State machine error in $fname" );
2353                         }
2354                 }
2355                 if( $stack > 0 ) {
2356                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2357                         return false;
2358                 }
2359                 wfProfileOut( $fname );
2360                 return false;
2361         }
2362
2363         /**
2364          * Return value of a magic variable (like PAGENAME)
2365          *
2366          * @private
2367          */
2368         function getVariableValue( $index ) {
2369                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2370
2371                 /**
2372                  * Some of these require message or data lookups and can be
2373                  * expensive to check many times.
2374                  */
2375                 static $varCache = array();
2376                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2377                         if ( isset( $varCache[$index] ) ) {
2378                                 return $varCache[$index];
2379                         }
2380                 }
2381
2382                 $ts = time();
2383                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2384
2385                 # Use the time zone
2386                 global $wgLocaltimezone;
2387                 if ( isset( $wgLocaltimezone ) ) {
2388                         $oldtz = getenv( 'TZ' );
2389                         putenv( 'TZ='.$wgLocaltimezone );
2390                 }
2391
2392                 wfSuppressWarnings(); // E_STRICT system time bitching
2393                 $localTimestamp = date( 'YmdHis', $ts );
2394                 $localMonth = date( 'm', $ts );
2395                 $localMonthName = date( 'n', $ts );
2396                 $localDay = date( 'j', $ts );
2397                 $localDay2 = date( 'd', $ts );
2398                 $localDayOfWeek = date( 'w', $ts );
2399                 $localWeek = date( 'W', $ts );
2400                 $localYear = date( 'Y', $ts );
2401                 $localHour = date( 'H', $ts );
2402                 if ( isset( $wgLocaltimezone ) ) {
2403                         putenv( 'TZ='.$oldtz );
2404                 }
2405                 wfRestoreWarnings();
2406
2407                 switch ( $index ) {
2408                         case 'currentmonth':
2409                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2410                         case 'currentmonthname':
2411                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2412                         case 'currentmonthnamegen':
2413                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2414                         case 'currentmonthabbrev':
2415                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2416                         case 'currentday':
2417                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2418                         case 'currentday2':
2419                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2420                         case 'localmonth':
2421                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2422                         case 'localmonthname':
2423                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2424                         case 'localmonthnamegen':
2425                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2426                         case 'localmonthabbrev':
2427                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2428                         case 'localday':
2429                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2430                         case 'localday2':
2431                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2432                         case 'pagename':
2433                                 return wfEscapeWikiText( $this->mTitle->getText() );
2434                         case 'pagenamee':
2435                                 return $this->mTitle->getPartialURL();
2436                         case 'fullpagename':
2437                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2438                         case 'fullpagenamee':
2439                                 return $this->mTitle->getPrefixedURL();
2440                         case 'subpagename':
2441                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2442                         case 'subpagenamee':
2443                                 return $this->mTitle->getSubpageUrlForm();
2444                         case 'basepagename':
2445                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2446                         case 'basepagenamee':
2447                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2448                         case 'talkpagename':
2449                                 if( $this->mTitle->canTalk() ) {
2450                                         $talkPage = $this->mTitle->getTalkPage();
2451                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2452                                 } else {
2453                                         return '';
2454                                 }
2455                         case 'talkpagenamee':
2456                                 if( $this->mTitle->canTalk() ) {
2457                                         $talkPage = $this->mTitle->getTalkPage();
2458                                         return $talkPage->getPrefixedUrl();
2459                                 } else {
2460                                         return '';
2461                                 }
2462                         case 'subjectpagename':
2463                                 $subjPage = $this->mTitle->getSubjectPage();
2464                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2465                         case 'subjectpagenamee':
2466                                 $subjPage = $this->mTitle->getSubjectPage();
2467                                 return $subjPage->getPrefixedUrl();
2468                         case 'revisionid':
2469                                 return $this->mRevisionId;
2470                         case 'revisionday':
2471                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2472                         case 'revisionday2':
2473                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2474                         case 'revisionmonth':
2475                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2476                         case 'revisionyear':
2477                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2478                         case 'revisiontimestamp':
2479                                 return $this->getRevisionTimestamp();
2480                         case 'namespace':
2481                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2482                         case 'namespacee':
2483                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2484                         case 'talkspace':
2485                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2486                         case 'talkspacee':
2487                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2488                         case 'subjectspace':
2489                                 return $this->mTitle->getSubjectNsText();
2490                         case 'subjectspacee':
2491                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2492                         case 'currentdayname':
2493                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2494                         case 'currentyear':
2495                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2496                         case 'currenttime':
2497                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2498                         case 'currenthour':
2499                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2500                         case 'currentweek':
2501                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2502                                 // int to remove the padding
2503                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2504                         case 'currentdow':
2505                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2506                         case 'localdayname':
2507                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2508                         case 'localyear':
2509                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2510                         case 'localtime':
2511                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2512                         case 'localhour':
2513                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2514                         case 'localweek':
2515                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2516                                 // int to remove the padding
2517                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2518                         case 'localdow':
2519                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2520                         case 'numberofarticles':
2521                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2522                         case 'numberoffiles':
2523                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2524                         case 'numberofusers':
2525                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2526                         case 'numberofpages':
2527                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2528                         case 'numberofadmins':
2529                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2530                         case 'numberofedits':
2531                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2532                         case 'currenttimestamp':
2533                                 return $varCache[$index] = wfTimestampNow();
2534                         case 'localtimestamp':
2535                                 return $varCache[$index] = $localTimestamp;
2536                         case 'currentversion':
2537                                 return $varCache[$index] = SpecialVersion::getVersion();
2538                         case 'sitename':
2539                                 return $wgSitename;
2540                         case 'server':
2541                                 return $wgServer;
2542                         case 'servername':
2543                                 return $wgServerName;
2544                         case 'scriptpath':
2545                                 return $wgScriptPath;
2546                         case 'directionmark':
2547                                 return $wgContLang->getDirMark();
2548                         case 'contentlanguage':
2549                                 global $wgContLanguageCode;
2550                                 return $wgContLanguageCode;
2551                         default:
2552                                 $ret = null;
2553                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2554                                         return $ret;
2555                                 else
2556                                         return null;
2557                 }
2558         }
2559
2560         /**
2561          * initialise the magic variables (like CURRENTMONTHNAME)
2562          *
2563          * @private
2564          */
2565         function initialiseVariables() {
2566                 $fname = 'Parser::initialiseVariables';
2567                 wfProfileIn( $fname );
2568                 $variableIDs = MagicWord::getVariableIDs();
2569
2570                 $this->mVariables = array();
2571                 foreach ( $variableIDs as $id ) {
2572                         $mw =& MagicWord::get( $id );
2573                         $mw->addToArray( $this->mVariables, $id );
2574                 }
2575                 wfProfileOut( $fname );
2576         }
2577
2578         /**
2579          * parse any parentheses in format ((title|part|part))
2580          * and call callbacks to get a replacement text for any found piece
2581          *
2582          * @param string $text The text to parse
2583          * @param array $callbacks rules in form:
2584          *     '{' => array(                            # opening parentheses
2585          *                                      'end' => '}',   # closing parentheses
2586          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2587          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2588          *                                                                )
2589          *                                      )
2590          *                                      'min' => 2,     # Minimum parenthesis count in cb
2591          *                                      'max' => 3,     # Maximum parenthesis count in cb
2592          * @private
2593          */
2594         function replace_callback ($text, $callbacks) {
2595                 wfProfileIn( __METHOD__ );
2596                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2597                 $lastOpeningBrace = -1;                 # last not closed parentheses
2598
2599                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2600
2601                 $i = 0;
2602                 while ( $i < strlen( $text ) ) {
2603                         # Find next opening brace, closing brace or pipe
2604                         if ( $lastOpeningBrace == -1 ) {
2605                                 $currentClosing = '';
2606                                 $search = $validOpeningBraces;
2607                         } else {
2608                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2609                                 $search = $validOpeningBraces . '|' . $currentClosing;
2610                         }
2611                         $rule = null;
2612                         $i += strcspn( $text, $search, $i );
2613                         if ( $i < strlen( $text ) ) {
2614                                 if ( $text[$i] == '|' ) {
2615                                         $found = 'pipe';
2616                                 } elseif ( $text[$i] == $currentClosing ) {
2617                                         $found = 'close';
2618                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2619                                         $found = 'open';
2620                                         $rule = $callbacks[$text[$i]];
2621                                 } else {
2622                                         # Some versions of PHP have a strcspn which stops on null characters
2623                                         # Ignore and continue
2624                                         ++$i;
2625                                         continue;
2626                                 }
2627                         } else {
2628                                 # All done
2629                                 break;
2630                         }
2631
2632                         if ( $found == 'open' ) {
2633                                 # found opening brace, let's add it to parentheses stack
2634                                 $piece = array('brace' => $text[$i],
2635                                                            'braceEnd' => $rule['end'],
2636                                                            'title' => '',
2637                                                            'parts' => null);
2638
2639                                 # count opening brace characters
2640                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2641                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2642                                 $i += $piece['count'];
2643
2644                                 # we need to add to stack only if opening brace count is enough for one of the rules
2645                                 if ( $piece['count'] >= $rule['min'] ) {
2646                                         $lastOpeningBrace ++;
2647                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2648                                 }
2649                         } elseif ( $found == 'close' ) {
2650                                 # lets check if it is enough characters for closing brace
2651                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2652                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2653
2654                                 # check for maximum matching characters (if there are 5 closing
2655                                 # characters, we will probably need only 3 - depending on the rules)
2656                                 $matchingCount = 0;
2657                                 $matchingCallback = null;
2658                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2659                                 if ( $count > $cbType['max'] ) {
2660                                         # The specified maximum exists in the callback array, unless the caller
2661                                         # has made an error
2662                                         $matchingCount = $cbType['max'];
2663                                 } else {
2664                                         # Count is less than the maximum
2665                                         # Skip any gaps in the callback array to find the true largest match
2666                                         # Need to use array_key_exists not isset because the callback can be null
2667                                         $matchingCount = $count;
2668                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2669                                                 --$matchingCount;
2670                                         }
2671                                 }
2672
2673                                 if ($matchingCount <= 0) {
2674                                         $i += $count;
2675                                         continue;
2676                                 }
2677                                 $matchingCallback = $cbType['cb'][$matchingCount];
2678
2679                                 # let's set a title or last part (if '|' was found)
2680                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2681                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2682                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2683                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2684                                 } else {
2685                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2686                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2687                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2688                                 }
2689
2690                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2691                                 $pieceEnd = $i + $matchingCount;
2692
2693                                 if( is_callable( $matchingCallback ) ) {
2694                                         $cbArgs = array (
2695                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2696                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2697                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2698                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2699                                                                          );
2700                                         # finally we can call a user callback and replace piece of text
2701                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2702                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2703                                         $i = $pieceStart + strlen($replaceWith);
2704                                 } else {
2705                                         # null value for callback means that parentheses should be parsed, but not replaced
2706                                         $i += $matchingCount;
2707                                 }
2708
2709                                 # reset last opening parentheses, but keep it in case there are unused characters
2710                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2711                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2712                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2713                                                            'title' => '',
2714                                                            'parts' => null,
2715                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2716                                 $openingBraceStack[$lastOpeningBrace--] = null;
2717
2718                                 if ($matchingCount < $piece['count']) {
2719                                         $piece['count'] -= $matchingCount;
2720                                         $piece['startAt'] -= $matchingCount;
2721                                         $piece['partStart'] = $piece['startAt'];
2722                                         # do we still qualify for any callback with remaining count?
2723                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2724                                         while ( $piece['count'] ) {
2725                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2726                                                         $lastOpeningBrace++;
2727                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2728                                                         break;
2729                                                 }
2730                                                 --$piece['count'];
2731                                         }
2732                                 }
2733                         } elseif ( $found == 'pipe' ) {
2734                                 # lets set a title if it is a first separator, or next part otherwise
2735                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2736                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2737                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2738                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2739                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2740                                 } else {
2741                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2742                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2743                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2744                                 }
2745                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2746                         }
2747                 }
2748
2749                 wfProfileOut( __METHOD__ );
2750                 return $text;
2751         }
2752
2753         /**
2754          * Replace magic variables, templates, and template arguments
2755          * with the appropriate text. Templates are substituted recursively,
2756          * taking care to avoid infinite loops.
2757          *
2758          * Note that the substitution depends on value of $mOutputType:
2759          *  OT_WIKI: only {{subst:}} templates
2760          *  OT_MSG: only magic variables
2761          *  OT_HTML: all templates and magic variables
2762          *
2763          * @param string $tex The text to transform
2764          * @param array $args Key-value pairs representing template parameters to substitute
2765          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2766          * @private
2767          */
2768         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2769                 # Prevent too big inclusions
2770                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2771                         return $text;
2772                 }
2773
2774                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2775                 wfProfileIn( $fname );
2776
2777                 # This function is called recursively. To keep track of arguments we need a stack:
2778                 array_push( $this->mArgStack, $args );
2779
2780                 $braceCallbacks = array();
2781                 if ( !$argsOnly ) {
2782                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2783                 }
2784                 if ( $this->mOutputType != OT_MSG ) {
2785                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2786                 }
2787                 if ( $braceCallbacks ) {
2788                         $callbacks = array(
2789                                 '{' => array(
2790                                         'end' => '}',
2791                                         'cb' => $braceCallbacks,
2792                                         'min' => $argsOnly ? 3 : 2,
2793                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2794                                 ),
2795                                 '[' => array(
2796                                         'end' => ']',
2797                                         'cb' => array(2=>null),
2798                                         'min' => 2,
2799                                         'max' => 2,
2800                                 )
2801                         );
2802                         $text = $this->replace_callback ($text, $callbacks);
2803
2804                         array_pop( $this->mArgStack );
2805                 }
2806                 wfProfileOut( $fname );
2807                 return $text;
2808         }
2809
2810         /**
2811          * Replace magic variables
2812          * @private
2813          */
2814         function variableSubstitution( $matches ) {
2815                 global $wgContLang;
2816                 $fname = 'Parser::variableSubstitution';
2817                 $varname = $wgContLang->lc($matches[1]);
2818                 wfProfileIn( $fname );
2819                 $skip = false;
2820                 if ( $this->mOutputType == OT_WIKI ) {
2821                         # Do only magic variables prefixed by SUBST
2822                         $mwSubst =& MagicWord::get( 'subst' );
2823                         if (!$mwSubst->matchStartAndRemove( $varname ))
2824                                 $skip = true;
2825                         # Note that if we don't substitute the variable below,
2826                         # we don't remove the {{subst:}} magic word, in case
2827                         # it is a template rather than a magic variable.
2828                 }
2829                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2830                         $id = $this->mVariables[$varname];
2831                         # Now check if we did really match, case sensitive or not
2832                         $mw =& MagicWord::get( $id );
2833                         if ($mw->match($matches[1])) {
2834                                 $text = $this->getVariableValue( $id );
2835                                 $this->mOutput->mContainsOldMagic = true;
2836                         } else {
2837                                 $text = $matches[0];
2838                         }
2839                 } else {
2840                         $text = $matches[0];
2841                 }
2842                 wfProfileOut( $fname );
2843                 return $text;
2844         }
2845
2846
2847         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2848         static function createAssocArgs( $args ) {
2849                 $assocArgs = array();
2850                 $index = 1;
2851                 foreach( $args as $arg ) {
2852                         $eqpos = strpos( $arg, '=' );
2853                         if ( $eqpos === false ) {
2854                                 $assocArgs[$index++] = $arg;
2855                         } else {
2856                                 $name = trim( substr( $arg, 0, $eqpos ) );
2857                                 $value = trim( substr( $arg, $eqpos+1 ) );
2858                                 if ( $value === false ) {
2859                                         $value = '';
2860                                 }
2861                                 if ( $name !== false ) {
2862                                         $assocArgs[$name] = $value;
2863                                 }
2864                         }
2865                 }
2866
2867                 return $assocArgs;
2868         }
2869
2870         /**
2871          * Return the text of a template, after recursively
2872          * replacing any variables or templates within the template.
2873          *
2874          * @param array $piece The parts of the template
2875          *  $piece['text']: matched text
2876          *  $piece['title']: the title, i.e. the part before the |
2877          *  $piece['parts']: the parameter array
2878          * @return string the text of the template
2879          * @private
2880          */
2881         function braceSubstitution( $piece ) {
2882                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2883                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2884                 wfProfileIn( $fname );
2885                 wfProfileIn( __METHOD__.'-setup' );
2886
2887                 # Flags
2888                 $found = false;             # $text has been filled
2889                 $nowiki = false;            # wiki markup in $text should be escaped
2890                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2891                 $noargs = false;            # Don't replace triple-brace arguments in $text
2892                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2893                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2894                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2895                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2896
2897                 # Title object, where $text came from
2898                 $title = NULL;
2899
2900                 $linestart = '';
2901
2902
2903                 # $part1 is the bit before the first |, and must contain only title characters
2904                 # $args is a list of arguments, starting from index 0, not including $part1
2905
2906                 $titleText = $part1 = $piece['title'];
2907                 # If the third subpattern matched anything, it will start with |
2908
2909                 if (null == $piece['parts']) {
2910                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2911                         if ($replaceWith != $piece['text']) {
2912                                 $text = $replaceWith;
2913                                 $found = true;
2914                                 $noparse = true;
2915                                 $noargs = true;
2916                         }
2917                 }
2918
2919                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2920                 wfProfileOut( __METHOD__.'-setup' );
2921
2922                 # SUBST
2923                 wfProfileIn( __METHOD__.'-modifiers' );
2924                 if ( !$found ) {
2925                         $mwSubst =& MagicWord::get( 'subst' );
2926                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2927                                 # One of two possibilities is true:
2928                                 # 1) Found SUBST but not in the PST phase
2929                                 # 2) Didn't find SUBST and in the PST phase
2930                                 # In either case, return without further processing
2931                                 $text = $piece['text'];
2932                                 $found = true;
2933                                 $noparse = true;
2934                                 $noargs = true;
2935                         }
2936                 }
2937
2938                 # MSG, MSGNW and RAW
2939                 if ( !$found ) {
2940                         # Check for MSGNW:
2941                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2942                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2943                                 $nowiki = true;
2944                         } else {
2945                                 # Remove obsolete MSG:
2946                                 $mwMsg =& MagicWord::get( 'msg' );
2947                                 $mwMsg->matchStartAndRemove( $part1 );
2948                         }
2949
2950                         # Check for RAW:
2951                         $mwRaw =& MagicWord::get( 'raw' );
2952                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2953                                 $forceRawInterwiki = true;
2954                         }
2955                 }
2956                 wfProfileOut( __METHOD__.'-modifiers' );
2957
2958                 //save path level before recursing into functions & templates.
2959                 $lastPathLevel = $this->mTemplatePath;
2960
2961                 # Parser functions
2962                 if ( !$found ) {
2963                         wfProfileIn( __METHOD__ . '-pfunc' );
2964
2965                         $colonPos = strpos( $part1, ':' );
2966                         if ( $colonPos !== false ) {
2967                                 # Case sensitive functions
2968                                 $function = substr( $part1, 0, $colonPos );
2969                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2970                                         $function = $this->mFunctionSynonyms[1][$function];
2971                                 } else {
2972                                         # Case insensitive functions
2973                                         $function = strtolower( $function );
2974                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2975                                                 $function = $this->mFunctionSynonyms[0][$function];
2976                                         } else {
2977                                                 $function = false;
2978                                         }
2979                                 }
2980                                 if ( $function ) {
2981                                         $funcArgs = array_map( 'trim', $args );
2982                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2983                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2984                                         $found = true;
2985
2986                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2987                                         //$noargs = true;
2988                                         //$noparse = true;
2989
2990                                         if ( is_array( $result ) ) {
2991                                                 if ( isset( $result[0] ) ) {
2992                                                         $text = $linestart . $result[0];
2993                                                         unset( $result[0] );
2994                                                 }
2995
2996                                                 // Extract flags into the local scope
2997                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2998                                                 extract( $result );
2999                                         } else {
3000                                                 $text = $linestart . $result;
3001                                         }
3002                                 }
3003                         }
3004                         wfProfileOut( __METHOD__ . '-pfunc' );
3005                 }
3006
3007                 # Template table test
3008
3009                 # Did we encounter this template already? If yes, it is in the cache
3010                 # and we need to check for loops.
3011                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3012                         $found = true;
3013
3014                         # Infinite loop test
3015                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3016                                 $noparse = true;
3017                                 $noargs = true;
3018                                 $found = true;
3019                                 $text = $linestart .
3020                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3021                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3022                         } else {
3023                                 # set $text to cached message.
3024                                 $text = $linestart . $this->mTemplates[$piece['title']];
3025                                 #treat title for cached page the same as others
3026                                 $ns = NS_TEMPLATE;
3027                                 $subpage = '';
3028                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3029                                 if ($subpage !== '') {
3030                                   $ns = $this->mTitle->getNamespace();
3031                                 }
3032                                 $title = Title::newFromText( $part1, $ns );
3033                                 //used by include size checking
3034                                 $titleText = $title->getPrefixedText();
3035                                 //used by edit section links
3036                                 $replaceHeadings = true;
3037
3038                         }
3039                 }
3040
3041                 # Load from database
3042                 if ( !$found ) {
3043                         wfProfileIn( __METHOD__ . '-loadtpl' );
3044                         $ns = NS_TEMPLATE;
3045                         # declaring $subpage directly in the function call
3046                         # does not work correctly with references and breaks
3047                         # {{/subpage}}-style inclusions
3048                         $subpage = '';
3049                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3050                         if ($subpage !== '') {
3051                                 $ns = $this->mTitle->getNamespace();
3052                         }
3053                         $title = Title::newFromText( $part1, $ns );
3054
3055
3056                         if ( !is_null( $title ) ) {
3057                                 $titleText = $title->getPrefixedText();
3058                                 # Check for language variants if the template is not found
3059                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3060                                         $wgContLang->findVariantLink($part1, $title);
3061                                 }
3062
3063                                 if ( !$title->isExternal() ) {
3064                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3065                                                 $text = SpecialPage::capturePath( $title );
3066                                                 if ( is_string( $text ) ) {
3067                                                         $found = true;
3068                                                         $noparse = true;
3069                                                         $noargs = true;
3070                                                         $isHTML = true;
3071                                                         $this->disableCache();
3072                                                 }
3073                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3074                                                 $found = false; //access denied
3075                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3076                                         } else {
3077                                                 list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3078                                                 if ( $articleContent !== false ) {
3079                                                         $found = true;
3080                                                         $text = $articleContent;
3081                                                         $replaceHeadings = true;
3082                                                 }
3083                                         }
3084
3085                                         # If the title is valid but undisplayable, make a link to it
3086                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3087                                                 $text = "[[:$titleText]]";
3088                                                 $found = true;
3089                                         }
3090                                 } elseif ( $title->isTrans() ) {
3091                                         // Interwiki transclusion
3092                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3093                                                 $text = $this->interwikiTransclude( $title, 'render' );
3094                                                 $isHTML = true;
3095                                                 $noparse = true;
3096                                         } else {
3097                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3098                                                 $replaceHeadings = true;
3099                                         }
3100                                         $found = true;
3101                                 }
3102
3103                                 # Template cache array insertion
3104                                 # Use the original $piece['title'] not the mangled $part1, so that
3105                                 # modifiers such as RAW: produce separate cache entries
3106                                 if( $found ) {
3107                                         if( $isHTML ) {
3108                                                 // A special page; don't store it in the template cache.
3109                                         } else {
3110                                                 $this->mTemplates[$piece['title']] = $text;
3111                                         }
3112                                         $text = $linestart . $text;
3113                                 }
3114                         }
3115                         wfProfileOut( __METHOD__ . '-loadtpl' );
3116                 }
3117
3118                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3119                         # Error, oversize inclusion
3120                         $text = $linestart .
3121                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3122                         $noparse = true;
3123                         $noargs = true;
3124                 }
3125
3126                 # Recursive parsing, escaping and link table handling
3127                 # Only for HTML output
3128                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3129                         $text = wfEscapeWikiText( $text );
3130                 } elseif ( !$this->ot['msg'] && $found ) {
3131                         if ( $noargs ) {
3132                                 $assocArgs = array();
3133                         } else {
3134                                 # Clean up argument array
3135                                 $assocArgs = self::createAssocArgs($args);
3136                                 # Add a new element to the templace recursion path
3137                                 $this->mTemplatePath[$part1] = 1;
3138                         }
3139
3140                         if ( !$noparse ) {
3141                                 # If there are any <onlyinclude> tags, only include them
3142                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3143                                         $replacer = new OnlyIncludeReplacer;
3144                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3145                                                 array( &$replacer, 'replace' ), $text );
3146                                         $text = $replacer->output;
3147                                 }
3148                                 # Remove <noinclude> sections and <includeonly> tags
3149                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3150                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3151
3152                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3153                                         # Strip <nowiki>, <pre>, etc.
3154                                         $text = $this->strip( $text, $this->mStripState );
3155                                         if ( $this->ot['html'] ) {
3156                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3157                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3158                                                 $text = Sanitizer::removeHTMLcomments( $text );
3159                                         }
3160                                 }
3161                                 $text = $this->replaceVariables( $text, $assocArgs );
3162
3163                                 # If the template begins with a table or block-level
3164                                 # element, it should be treated as beginning a new line.
3165                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3166                                         $text = "\n" . $text;
3167                                 }
3168                         } elseif ( !$noargs ) {
3169                                 # $noparse and !$noargs
3170                                 # Just replace the arguments, not any double-brace items
3171                                 # This is used for rendered interwiki transclusion
3172                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3173                         }
3174                 }
3175                 # Prune lower levels off the recursion check path
3176                 $this->mTemplatePath = $lastPathLevel;
3177
3178                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3179                         # Error, oversize inclusion
3180                         $text = $linestart .
3181                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3182                         $noparse = true;
3183                         $noargs = true;
3184                 }
3185
3186                 if ( !$found ) {
3187                         wfProfileOut( $fname );
3188                         return $piece['text'];
3189                 } else {
3190                         wfProfileIn( __METHOD__ . '-placeholders' );
3191                         if ( $isHTML ) {
3192                                 # Replace raw HTML by a placeholder
3193                                 # Add a blank line preceding, to prevent it from mucking up
3194                                 # immediately preceding headings
3195                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3196                         } else {
3197                                 # replace ==section headers==
3198                                 # XXX this needs to go away once we have a better parser.
3199                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3200                                         if( !is_null( $title ) )
3201                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3202                                         else
3203                                                 $encodedname = base64_encode("");
3204                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3205                                                 PREG_SPLIT_DELIM_CAPTURE);
3206                                         $text = '';
3207                                         $nsec = $headingOffset;
3208
3209                                         for( $i = 0; $i < count($m); $i += 2 ) {
3210                                                 $text .= $m[$i];
3211                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3212                                                 $hl = $m[$i + 1];
3213                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3214                                                         $text .= $hl;
3215                                                         continue;
3216                                                 }
3217                                                 $m2 = array();
3218                                                 preg_match('/^(={1,6})(.*?)(={1,6}\s*?)$/m', $hl, $m2);
3219                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3220                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3221
3222                                                 $nsec++;
3223                                         }
3224                                 }
3225                         }
3226                         wfProfileOut( __METHOD__ . '-placeholders' );
3227                 }
3228
3229                 # Prune lower levels off the recursion check path
3230                 $this->mTemplatePath = $lastPathLevel;
3231
3232                 if ( !$found ) {
3233                         wfProfileOut( $fname );
3234                         return $piece['text'];
3235                 } else {
3236                         wfProfileOut( $fname );
3237                         return $text;
3238                 }
3239         }
3240
3241         /**
3242          * Fetch the unparsed text of a template and register a reference to it.
3243          */
3244         function fetchTemplateAndTitle( $title ) {
3245                 $templateCb = $this->mOptions->getTemplateCallback();
3246                 $stuff = call_user_func( $templateCb, $title );
3247                 $text = $stuff['text'];
3248                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3249                 if ( isset( $stuff['deps'] ) ) {
3250                         foreach ( $stuff['deps'] as $dep ) {
3251                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3252                         }
3253                 }
3254                 return array($text,$finalTitle);
3255         }
3256
3257         function fetchTemplate( $title ) {
3258                 $rv = $this->fetchTemplateAndtitle($title);
3259                 return $rv[0];
3260         }
3261
3262         /**
3263          * Static function to get a template
3264          * Can be overridden via ParserOptions::setTemplateCallback().
3265          *
3266          * Returns an associative array:
3267          *    text          The unparsed template text
3268          *    finalTitle    (Optional) The title after following redirects
3269          *    deps          (Optional) An array of associative array dependencies:
3270          *                       title:    The dependency title, to be registered in templatelinks
3271          *                       page_id:  The page_id of the title
3272          *                       rev_id:   The revision ID loaded
3273          */
3274         static function statelessFetchTemplate( $title ) {
3275                 $text = $skip = false;
3276                 $finalTitle = $title;
3277                 $deps = array();
3278
3279                 // Loop to fetch the article, with up to 1 redirect
3280                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3281                         # Give extensions a chance to select the revision instead
3282                         $id = false; // Assume current
3283                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3284
3285                         if( $skip ) {
3286                                 $text = false;
3287                                 $deps[] = array(
3288                                         'title' => $title,
3289                                         'page_id' => $title->getArticleID(),
3290                                         'rev_id' => null );
3291                                 break;
3292                         }
3293                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3294                         $rev_id = $rev ? $rev->getId() : 0;
3295
3296                         $deps[] = array(
3297                                 'title' => $title,
3298                                 'page_id' => $title->getArticleID(),
3299                                 'rev_id' => $rev_id );
3300
3301                         if( $rev ) {
3302                                 $text = $rev->getText();
3303                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3304                                 global $wgLang;
3305                                 $message = $wgLang->lcfirst( $title->getText() );
3306                                 $text = wfMsgForContentNoTrans( $message );
3307                                 if( wfEmptyMsg( $message, $text ) ) {
3308                                         $text = false;
3309                                         break;
3310                                 }
3311                         } else {
3312                                 break;
3313                         }
3314                         if ( $text === false ) {
3315                                 break;
3316                         }
3317                         // Redirect?
3318                         $finalTitle = $title;
3319                         $title = Title::newFromRedirect( $text );
3320                 }
3321                 return array(
3322                         'text' => $text,
3323                         'finalTitle' => $finalTitle,
3324                         'deps' => $deps );
3325         }
3326
3327         /**
3328          * Transclude an interwiki link.
3329          */
3330         function interwikiTransclude( $title, $action ) {
3331                 global $wgEnableScaryTranscluding;
3332
3333                 if (!$wgEnableScaryTranscluding)
3334                         return wfMsg('scarytranscludedisabled');
3335
3336                 $url = $title->getFullUrl( "action=$action" );
3337
3338                 if (strlen($url) > 255)
3339                         return wfMsg('scarytranscludetoolong');
3340                 return $this->fetchScaryTemplateMaybeFromCache($url);
3341         }
3342
3343         function fetchScaryTemplateMaybeFromCache($url) {
3344                 global $wgTranscludeCacheExpiry;
3345                 $dbr = wfGetDB(DB_SLAVE);
3346                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3347                                 array('tc_url' => $url));
3348                 if ($obj) {
3349                         $time = $obj->tc_time;
3350                         $text = $obj->tc_contents;
3351                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3352                                 return $text;
3353                         }
3354                 }
3355
3356                 $text = Http::get($url);
3357                 if (!$text)
3358                         return wfMsg('scarytranscludefailed', $url);
3359
3360                 $dbw = wfGetDB(DB_MASTER);
3361                 $dbw->replace('transcache', array('tc_url'), array(
3362                         'tc_url' => $url,
3363                         'tc_time' => time(),
3364                         'tc_contents' => $text));
3365                 return $text;
3366         }
3367
3368
3369         /**
3370          * Triple brace replacement -- used for template arguments
3371          * @private
3372          */
3373         function argSubstitution( $matches ) {
3374                 $arg = trim( $matches['title'] );
3375                 $text = $matches['text'];
3376                 $inputArgs = end( $this->mArgStack );
3377
3378                 if ( array_key_exists( $arg, $inputArgs ) ) {
3379                         $text = $inputArgs[$arg];
3380                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3381                 null != $matches['parts'] && count($matches['parts']) > 0) {
3382                         $text = $matches['parts'][0];
3383                 }
3384                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3385                         $text = $matches['text'] .
3386                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3387                 }
3388
3389                 return $text;
3390         }
3391
3392         /**
3393          * Increment an include size counter
3394          *
3395          * @param string $type The type of expansion
3396          * @param integer $size The size of the text
3397          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3398          */
3399         function incrementIncludeSize( $type, $size ) {
3400                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3401                         return false;
3402                 } else {
3403                         $this->mIncludeSizes[$type] += $size;
3404                         return true;
3405                 }
3406         }
3407
3408         /**
3409          * Detect __NOGALLERY__ magic word and set a placeholder
3410          */
3411         function stripNoGallery( &$text ) {
3412                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3413                 # do not add TOC
3414                 $mw = MagicWord::get( 'nogallery' );
3415                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3416         }
3417
3418         /**
3419          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3420          * placeholder that will then be replaced by the real TOC in
3421          * ->formatHeadings, this works because at this points real
3422          * comments will have already been discarded by the sanitizer.
3423          *
3424          * Any additional __TOC__ magic words left over will be discarded
3425          * as there can only be one TOC on the page.
3426          */
3427         function stripToc( $text ) {
3428                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3429                 # do not add TOC
3430                 $mw = MagicWord::get( 'notoc' );
3431                 if( $mw->matchAndRemove( $text ) ) {
3432                         $this->mShowToc = false;
3433                 }
3434
3435                 $mw = MagicWord::get( 'toc' );
3436                 if( $mw->match( $text ) ) {
3437                         $this->mShowToc = true;
3438                         $this->mForceTocPosition = true;
3439
3440                         // Set a placeholder. At the end we'll fill it in with the TOC.
3441                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3442
3443                         // Only keep the first one.
3444                         $text = $mw->replace( '', $text );
3445                 }
3446                 return $text;
3447         }
3448
3449         /**
3450          * This function accomplishes several tasks:
3451          * 1) Auto-number headings if that option is enabled
3452          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3453          * 3) Add a Table of contents on the top for users who have enabled the option
3454          * 4) Auto-anchor headings
3455          *
3456          * It loops through all headlines, collects the necessary data, then splits up the
3457          * string and re-inserts the newly formatted headlines.
3458          *
3459          * @param string $text
3460          * @param boolean $isMain
3461          * @private
3462          */
3463         function formatHeadings( $text, $isMain=true ) {
3464                 global $wgMaxTocLevel, $wgContLang;
3465
3466                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3467                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3468                         $showEditLink = 0;
3469                 } else {
3470                         $showEditLink = $this->mOptions->getEditSection();
3471                 }
3472
3473                 # Inhibit editsection links if requested in the page
3474                 $esw =& MagicWord::get( 'noeditsection' );
3475                 if( $esw->matchAndRemove( $text ) ) {
3476                         $showEditLink = 0;
3477                 }
3478
3479                 # Get all headlines for numbering them and adding funky stuff like [edit]
3480                 # links - this is for later, but we need the number of headlines right now
3481                 $matches = array();
3482                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3483
3484                 # if there are fewer than 4 headlines in the article, do not show TOC
3485                 # unless it's been explicitly enabled.
3486                 $enoughToc = $this->mShowToc &&
3487                         (($numMatches >= 4) || $this->mForceTocPosition);
3488
3489                 # Allow user to stipulate that a page should have a "new section"
3490                 # link added via __NEWSECTIONLINK__
3491                 $mw =& MagicWord::get( 'newsectionlink' );
3492                 if( $mw->matchAndRemove( $text ) )
3493                         $this->mOutput->setNewSection( true );
3494
3495                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3496                 # override above conditions and always show TOC above first header
3497                 $mw =& MagicWord::get( 'forcetoc' );
3498                 if ($mw->matchAndRemove( $text ) ) {
3499                         $this->mShowToc = true;
3500                         $enoughToc = true;
3501                 }
3502
3503                 # We need this to perform operations on the HTML
3504                 $sk = $this->mOptions->getSkin();
3505
3506                 # headline counter
3507                 $headlineCount = 0;
3508                 $sectionCount = 0; # headlineCount excluding template sections
3509                 $numVisible = 0;
3510
3511                 # Ugh .. the TOC should have neat indentation levels which can be
3512                 # passed to the skin functions. These are determined here
3513                 $toc = '';
3514                 $full = '';
3515                 $head = array();
3516                 $sublevelCount = array();
3517                 $levelCount = array();
3518                 $toclevel = 0;
3519                 $level = 0;
3520                 $prevlevel = 0;
3521                 $toclevel = 0;
3522                 $prevtoclevel = 0;
3523                 $rawtoc = array();
3524
3525                 foreach( $matches[3] as $headline ) {
3526                         $istemplate = 0;
3527                         $templatetitle = '';
3528                         $templatesection = 0;
3529                         $numbering = '';
3530                         $mat = array();
3531                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3532                                 $istemplate = 1;
3533                                 $templatetitle = base64_decode($mat[1]);
3534                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3535                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3536                         }
3537
3538                         if( $toclevel ) {
3539                                 $prevlevel = $level;
3540                                 $prevtoclevel = $toclevel;
3541                         }
3542                         $level = $matches[1][$headlineCount];
3543
3544                         if( $doNumberHeadings || $enoughToc ) {
3545
3546                                 if ( $level > $prevlevel ) {
3547                                         # Increase TOC level
3548                                         $toclevel++;
3549                                         $sublevelCount[$toclevel] = 0;
3550                                         if( $toclevel<$wgMaxTocLevel ) {
3551                                                 $prevtoclevel = $toclevel;
3552                                                 $toc .= $sk->tocIndent();
3553                                                 $numVisible++;
3554                                         }
3555                                 }
3556                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3557                                         # Decrease TOC level, find level to jump to
3558
3559                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3560                                                 # Can only go down to level 1
3561                                                 $toclevel = 1;
3562                                         } else {
3563                                                 for ($i = $toclevel; $i > 0; $i--) {
3564                                                         if ( $levelCount[$i] == $level ) {
3565                                                                 # Found last matching level
3566                                                                 $toclevel = $i;
3567                                                                 break;
3568                                                         }
3569                                                         elseif ( $levelCount[$i] < $level ) {
3570                                                                 # Found first matching level below current level
3571                                                                 $toclevel = $i + 1;
3572                                                                 break;
3573                                                         }
3574                                                 }
3575                                         }
3576                                         if( $toclevel<$wgMaxTocLevel ) {
3577                                                 if($prevtoclevel < $wgMaxTocLevel) {
3578                                                         # Unindent only if the previous toc level was shown :p
3579                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3580                                                 } else {
3581                                                         $toc .= $sk->tocLineEnd();
3582                                                 }
3583                                         }
3584                                 }
3585                                 else {
3586                                         # No change in level, end TOC line
3587                                         if( $toclevel<$wgMaxTocLevel ) {
3588                                                 $toc .= $sk->tocLineEnd();
3589                                         }
3590                                 }
3591
3592                                 $levelCount[$toclevel] = $level;
3593
3594                                 # count number of headlines for each level
3595                                 @$sublevelCount[$toclevel]++;
3596                                 $dot = 0;
3597                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3598                                         if( !empty( $sublevelCount[$i] ) ) {
3599                                                 if( $dot ) {
3600                                                         $numbering .= '.';
3601                                                 }
3602                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3603                                                 $dot = 1;
3604                                         }
3605                                 }
3606                         }
3607
3608                         # The canonized header is a version of the header text safe to use for links
3609                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3610                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3611
3612                         # Remove link placeholders by the link text.
3613                         #     <!--LINK number-->
3614                         # turns into
3615                         #     link text with suffix
3616                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3617                                                             "\$this->mLinkHolders['texts'][\$1]",
3618                                                             $canonized_headline );
3619                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3620                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3621                                                             $canonized_headline );
3622
3623                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3624                         $tocline = preg_replace(
3625                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3626                                 array( '',                          '<$1>'),
3627                                 $canonized_headline
3628                         );
3629                         $tocline = trim( $tocline );
3630
3631                         # For the anchor, strip out HTML-y stuff period
3632                         $canonized_headline = preg_replace( '/<.*?'.'>/', '', $canonized_headline );
3633                         $canonized_headline = trim( $canonized_headline );
3634
3635                         # Save headline for section edit hint before it's escaped
3636                         $headline_hint = $canonized_headline;
3637                         $canonized_headline = Sanitizer::escapeId( $canonized_headline );
3638                         $refers[$headlineCount] = $canonized_headline;
3639
3640                         # count how many in assoc. array so we can track dupes in anchors
3641                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3642                         $refcount[$headlineCount]=$refers[$canonized_headline];
3643
3644                         # Don't number the heading if it is the only one (looks silly)
3645                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3646                                 # the two are different if the line contains a link
3647                                 $headline=$numbering . ' ' . $headline;
3648                         }
3649
3650                         # Create the anchor for linking from the TOC to the section
3651                         $anchor = $canonized_headline;
3652                         if($refcount[$headlineCount] > 1 ) {
3653                                 $anchor .= '_' . $refcount[$headlineCount];
3654                         }
3655                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3656                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3657                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3658                         }
3659                         # give headline the correct <h#> tag
3660                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3661                                 if( $istemplate )
3662                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3663                                 else
3664                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3665                         } else {
3666                                 $editlink = '';
3667                         }
3668                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3669
3670                         $headlineCount++;
3671                         if( !$istemplate )
3672                                 $sectionCount++;
3673                 }
3674
3675                 $this->mOutput->setSections( $tocraw );
3676
3677                 # Never ever show TOC if no headers
3678                 if( $numVisible < 1 ) {
3679                         $enoughToc = false;
3680                 }
3681
3682                 if( $enoughToc ) {
3683                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3684                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3685                         }
3686                         $toc = $sk->tocList( $toc );
3687                 }
3688
3689                 # split up and insert constructed headlines
3690
3691                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3692                 $i = 0;
3693
3694                 foreach( $blocks as $block ) {
3695                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3696                                 # This is the [edit] link that appears for the top block of text when
3697                                 # section editing is enabled
3698
3699                                 # Disabled because it broke block formatting
3700                                 # For example, a bullet point in the top line
3701                                 # $full .= $sk->editSectionLink(0);
3702                         }
3703                         $full .= $block;
3704                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3705                                 # Top anchor now in skin
3706                                 $full = $full.$toc;
3707                         }
3708
3709                         if( !empty( $head[$i] ) ) {
3710                                 $full .= $head[$i];
3711                         }
3712                         $i++;
3713                 }
3714                 if( $this->mForceTocPosition ) {
3715                         return str_replace( '<!--MWTOC-->', $toc, $full );
3716                 } else {
3717                         return $full;
3718                 }
3719         }
3720
3721         /**
3722          * Transform wiki markup when saving a page by doing \r\n -> \n
3723          * conversion, substitting signatures, {{subst:}} templates, etc.
3724          *
3725          * @param string $text the text to transform
3726          * @param Title &$title the Title object for the current article
3727          * @param User &$user the User object describing the current user
3728          * @param ParserOptions $options parsing options
3729          * @param bool $clearState whether to clear the parser state first
3730          * @return string the altered wiki markup
3731          * @public
3732          */
3733         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3734                 $this->mOptions = $options;
3735                 $this->mTitle =& $title;
3736                 $this->setOutputType( OT_WIKI );
3737
3738                 if ( $clearState ) {
3739                         $this->clearState();
3740                 }
3741
3742                 $stripState = new StripState;
3743                 $pairs = array(
3744                         "\r\n" => "\n",
3745                 );
3746                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3747                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3748                 $text = $this->pstPass2( $text, $stripState, $user );
3749                 $text = $stripState->unstripBoth( $text );
3750                 return $text;
3751         }
3752
3753         /**
3754          * Pre-save transform helper function
3755          * @private
3756          */
3757         function pstPass2( $text, &$stripState, $user ) {
3758                 global $wgContLang, $wgLocaltimezone;
3759
3760                 /* Note: This is the timestamp saved as hardcoded wikitext to
3761                  * the database, we use $wgContLang here in order to give
3762                  * everyone the same signature and use the default one rather
3763                  * than the one selected in each user's preferences.
3764                  */
3765                 if ( isset( $wgLocaltimezone ) ) {
3766                         $oldtz = getenv( 'TZ' );
3767                         putenv( 'TZ='.$wgLocaltimezone );
3768                 }
3769                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3770                   ' (' . date( 'T' ) . ')';
3771                 if ( isset( $wgLocaltimezone ) ) {
3772                         putenv( 'TZ='.$oldtz );
3773                 }
3774
3775                 # Variable replacement
3776                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3777                 $text = $this->replaceVariables( $text );
3778
3779                 # Strip out <nowiki> etc. added via replaceVariables
3780                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3781
3782                 # Signatures
3783                 $sigText = $this->getUserSig( $user );
3784                 $text = strtr( $text, array(
3785                         '~~~~~' => $d,
3786                         '~~~~' => "$sigText $d",
3787                         '~~~' => $sigText
3788                 ) );
3789
3790                 # Context links: [[|name]] and [[name (context)|]]
3791                 #
3792                 global $wgLegalTitleChars;
3793                 $tc = "[$wgLegalTitleChars]";
3794                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3795
3796                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3797                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3798                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3799
3800                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3801                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3802                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3803
3804                 $t = $this->mTitle->getText();
3805                 $m = array();
3806                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3807                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3808                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3809                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3810                 } else {
3811                         # if there's no context, don't bother duplicating the title
3812                         $text = preg_replace( $p2, '[[\\1]]', $text );
3813                 }
3814
3815                 # Trim trailing whitespace
3816                 $text = rtrim( $text );
3817
3818                 return $text;
3819         }
3820
3821         /**
3822          * Fetch the user's signature text, if any, and normalize to
3823          * validated, ready-to-insert wikitext.
3824          *
3825          * @param User $user
3826          * @return string
3827          * @private
3828          */
3829         function getUserSig( &$user ) {
3830                 global $wgMaxSigChars;
3831
3832                 $username = $user->getName();
3833                 $nickname = $user->getOption( 'nickname' );
3834                 $nickname = $nickname === '' ? $username : $nickname;
3835
3836                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3837                         $nickname = $username;
3838                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3839                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3840                         # Sig. might contain markup; validate this
3841                         if( $this->validateSig( $nickname ) !== false ) {
3842                                 # Validated; clean up (if needed) and return it
3843                                 return $this->cleanSig( $nickname, true );
3844                         } else {
3845                                 # Failed to validate; fall back to the default
3846                                 $nickname = $username;
3847                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3848                         }
3849                 }
3850
3851                 // Make sure nickname doesnt get a sig in a sig
3852                 $nickname = $this->cleanSigInSig( $nickname );
3853
3854                 # If we're still here, make it a link to the user page
3855                 $userText = wfEscapeWikiText( $username );
3856                 $nickText = wfEscapeWikiText( $nickname );
3857                 if ( $user->isAnon() )  {
3858                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3859                 } else {
3860                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3861                 }
3862         }
3863
3864         /**
3865          * Check that the user's signature contains no bad XML
3866          *
3867          * @param string $text
3868          * @return mixed An expanded string, or false if invalid.
3869          */
3870         function validateSig( $text ) {
3871                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3872         }
3873
3874         /**
3875          * Clean up signature text
3876          *
3877          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3878          * 2) Substitute all transclusions
3879          *
3880          * @param string $text
3881          * @param $parsing Whether we're cleaning (preferences save) or parsing
3882          * @return string Signature text
3883          */
3884         function cleanSig( $text, $parsing = false ) {
3885                 global $wgTitle;
3886                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3887
3888                 $substWord = MagicWord::get( 'subst' );
3889                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3890                 $substText = '{{' . $substWord->getSynonym( 0 );
3891
3892                 $text = preg_replace( $substRegex, $substText, $text );
3893                 $text = $this->cleanSigInSig( $text );
3894                 $text = $this->replaceVariables( $text );
3895
3896                 $this->clearState();
3897                 return $text;
3898         }
3899
3900         /**
3901          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3902          * @param string $text
3903          * @return string Signature text with /~{3,5}/ removed
3904          */
3905         function cleanSigInSig( $text ) {
3906                 $text = preg_replace( '/~{3,5}/', '', $text );
3907                 return $text;
3908         }
3909
3910         /**
3911          * Set up some variables which are usually set up in parse()
3912          * so that an external function can call some class members with confidence
3913          * @public
3914          */
3915         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3916                 $this->mTitle =& $title;
3917                 $this->mOptions = $options;
3918                 $this->setOutputType( $outputType );
3919                 if ( $clearState ) {
3920                         $this->clearState();
3921                 }
3922         }
3923
3924         /**
3925          * Transform a MediaWiki message by replacing magic variables.
3926          *
3927          * @param string $text the text to transform
3928          * @param ParserOptions $options  options
3929          * @return string the text with variables substituted
3930          * @public
3931          */
3932         function transformMsg( $text, $options ) {
3933                 global $wgTitle;
3934                 static $executing = false;
3935
3936                 $fname = "Parser::transformMsg";
3937
3938                 # Guard against infinite recursion
3939                 if ( $executing ) {
3940                         return $text;
3941                 }
3942                 $executing = true;
3943
3944                 wfProfileIn($fname);
3945
3946                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3947                         $this->mTitle = $wgTitle;
3948                 } else {
3949                         $this->mTitle = Title::newFromText('msg');
3950                 }
3951                 $this->mOptions = $options;
3952                 $this->setOutputType( OT_MSG );
3953                 $this->clearState();
3954                 $text = $this->replaceVariables( $text );
3955
3956                 $executing = false;
3957                 wfProfileOut($fname);
3958                 return $text;
3959         }
3960
3961         /**
3962          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3963          * The callback should have the following form:
3964          *    function myParserHook( $text, $params, &$parser ) { ... }
3965          *
3966          * Transform and return $text. Use $parser for any required context, e.g. use
3967          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3968          *
3969          * @public
3970          *
3971          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3972          * @param mixed $callback The callback function (and object) to use for the tag
3973          *
3974          * @return The old value of the mTagHooks array associated with the hook
3975          */
3976         function setHook( $tag, $callback ) {
3977                 $tag = strtolower( $tag );
3978                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3979                 $this->mTagHooks[$tag] = $callback;
3980
3981                 return $oldVal;
3982         }
3983
3984         function setTransparentTagHook( $tag, $callback ) {
3985                 $tag = strtolower( $tag );
3986                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3987                 $this->mTransparentTagHooks[$tag] = $callback;
3988
3989                 return $oldVal;
3990         }
3991
3992         /**
3993          * Create a function, e.g. {{sum:1|2|3}}
3994          * The callback function should have the form:
3995          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3996          *
3997          * The callback may either return the text result of the function, or an array with the text
3998          * in element 0, and a number of flags in the other elements. The names of the flags are
3999          * specified in the keys. Valid flags are:
4000          *   found                     The text returned is valid, stop processing the template. This
4001          *                             is on by default.
4002          *   nowiki                    Wiki markup in the return value should be escaped
4003          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4004          *   noargs                    Don't replace triple-brace arguments in the return value
4005          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4006          *
4007          * @public
4008          *
4009          * @param string $id The magic word ID
4010          * @param mixed $callback The callback function (and object) to use
4011          * @param integer $flags a combination of the following flags:
4012          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4013          *
4014          * @return The old callback function for this name, if any
4015          */
4016         function setFunctionHook( $id, $callback, $flags = 0 ) {
4017                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
4018                 $this->mFunctionHooks[$id] = $callback;
4019
4020                 # Add to function cache
4021                 $mw = MagicWord::get( $id );
4022                 if( !$mw )
4023                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4024
4025                 $synonyms = $mw->getSynonyms();
4026                 $sensitive = intval( $mw->isCaseSensitive() );
4027
4028                 foreach ( $synonyms as $syn ) {
4029                         # Case
4030                         if ( !$sensitive ) {
4031                                 $syn = strtolower( $syn );
4032                         }
4033                         # Add leading hash
4034                         if ( !( $flags & SFH_NO_HASH ) ) {
4035                                 $syn = '#' . $syn;
4036                         }
4037                         # Remove trailing colon
4038                         if ( substr( $syn, -1, 1 ) == ':' ) {
4039                                 $syn = substr( $syn, 0, -1 );
4040                         }
4041                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4042                 }
4043                 return $oldVal;
4044         }
4045
4046         /**
4047          * Get all registered function hook identifiers
4048          *
4049          * @return array
4050          */
4051         function getFunctionHooks() {
4052                 return array_keys( $this->mFunctionHooks );
4053         }
4054
4055         /**
4056          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4057          * Placeholders created in Skin::makeLinkObj()
4058          * Returns an array of links found, indexed by PDBK:
4059          *  0 - broken
4060          *  1 - normal link
4061          *  2 - stub
4062          * $options is a bit field, RLH_FOR_UPDATE to select for update
4063          */
4064         function replaceLinkHolders( &$text, $options = 0 ) {
4065                 global $wgUser;
4066                 global $wgContLang;
4067
4068                 $fname = 'Parser::replaceLinkHolders';
4069                 wfProfileIn( $fname );
4070
4071                 $pdbks = array();
4072                 $colours = array();
4073                 $sk = $this->mOptions->getSkin();
4074                 $linkCache =& LinkCache::singleton();
4075
4076                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4077                         wfProfileIn( $fname.'-check' );
4078                         $dbr = wfGetDB( DB_SLAVE );
4079                         $page = $dbr->tableName( 'page' );
4080                         $threshold = $wgUser->getOption('stubthreshold');
4081
4082                         # Sort by namespace
4083                         asort( $this->mLinkHolders['namespaces'] );
4084
4085                         # Generate query
4086                         $query = false;
4087                         $current = null;
4088                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4089                                 # Make title object
4090                                 $title = $this->mLinkHolders['titles'][$key];
4091
4092                                 # Skip invalid entries.
4093                                 # Result will be ugly, but prevents crash.
4094                                 if ( is_null( $title ) ) {
4095                                         continue;
4096                                 }
4097                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4098
4099                                 # Check if it's a static known link, e.g. interwiki
4100                                 if ( $title->isAlwaysKnown() ) {
4101                                         $colours[$pdbk] = 1;
4102                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4103                                         $colours[$pdbk] = 1;
4104                                         $this->mOutput->addLink( $title, $id );
4105                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4106                                         $colours[$pdbk] = 0;
4107                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4108                                         $colours[$pdbk] = 0;
4109                                 } else {
4110                                         # Not in the link cache, add it to the query
4111                                         if ( !isset( $current ) ) {
4112                                                 $current = $ns;
4113                                                 $query =  "SELECT page_id, page_namespace, page_title";
4114                                                 if ( $threshold > 0 ) {
4115                                                         $query .= ', page_len, page_is_redirect';
4116                                                 }
4117                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4118                                         } elseif ( $current != $ns ) {
4119                                                 $current = $ns;
4120                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4121                                         } else {
4122                                                 $query .= ', ';
4123                                         }
4124
4125                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4126                                 }
4127                         }
4128                         if ( $query ) {
4129                                 $query .= '))';
4130                                 if ( $options & RLH_FOR_UPDATE ) {
4131                                         $query .= ' FOR UPDATE';
4132                                 }
4133
4134                                 $res = $dbr->query( $query, $fname );
4135
4136                                 # Fetch data and form into an associative array
4137                                 # non-existent = broken
4138                                 # 1 = known
4139                                 # 2 = stub
4140                                 while ( $s = $dbr->fetchObject($res) ) {
4141                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4142                                         $pdbk = $title->getPrefixedDBkey();
4143                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4144                                         $this->mOutput->addLink( $title, $s->page_id );
4145
4146                                         $colours[$pdbk] = ( $threshold == 0 || (
4147                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4148                                                                 $s->page_is_redirect ||
4149                                                                 !Namespace::isContent( $s->page_namespace ) )
4150                                                             ? 1 : 2 );
4151                                 }
4152                         }
4153                         wfProfileOut( $fname.'-check' );
4154
4155                         # Do a second query for different language variants of links and categories
4156                         if($wgContLang->hasVariants()){
4157                                 $linkBatch = new LinkBatch();
4158                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4159                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4160                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4161
4162                                 $categories = $this->mOutput->getCategoryLinks();
4163
4164                                 // Add variants of links to link batch
4165                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4166                                         $title = $this->mLinkHolders['titles'][$key];
4167                                         if ( is_null( $title ) )
4168                                                 continue;
4169
4170                                         $pdbk = $title->getPrefixedDBkey();
4171                                         $titleText = $title->getText();
4172
4173                                         // generate all variants of the link title text
4174                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4175
4176                                         // if link was not found (in first query), add all variants to query
4177                                         if ( !isset($colours[$pdbk]) ){
4178                                                 foreach($allTextVariants as $textVariant){
4179                                                         if($textVariant != $titleText){
4180                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4181                                                                 if(is_null($variantTitle)) continue;
4182                                                                 $linkBatch->addObj( $variantTitle );
4183                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4184                                                         }
4185                                                 }
4186                                         }
4187                                 }
4188
4189                                 // process categories, check if a category exists in some variant
4190                                 foreach( $categories as $category ){
4191                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4192                                         foreach($variants as $variant){
4193                                                 if($variant != $category){
4194                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4195                                                         if(is_null($variantTitle)) continue;
4196                                                         $linkBatch->addObj( $variantTitle );
4197                                                         $categoryMap[$variant] = $category;
4198                                                 }
4199                                         }
4200                                 }
4201
4202
4203                                 if(!$linkBatch->isEmpty()){
4204                                         // construct query
4205                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4206
4207                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4208                                         if ( $threshold > 0 ) {
4209                                                 $variantQuery .= ', page_len, page_is_redirect';
4210                                         }
4211
4212                                         $variantQuery .= " FROM $page WHERE $titleClause";
4213                                         if ( $options & RLH_FOR_UPDATE ) {
4214                                                 $variantQuery .= ' FOR UPDATE';
4215                                         }
4216
4217                                         $varRes = $dbr->query( $variantQuery, $fname );
4218
4219                                         // for each found variants, figure out link holders and replace
4220                                         while ( $s = $dbr->fetchObject($varRes) ) {
4221
4222                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4223                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4224                                                 $vardbk = $variantTitle->getDBkey();
4225
4226                                                 $holderKeys = array();
4227                                                 if(isset($variantMap[$varPdbk])){
4228                                                         $holderKeys = $variantMap[$varPdbk];
4229                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4230                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4231                                                 }
4232
4233                                                 // loop over link holders
4234                                                 foreach($holderKeys as $key){
4235                                                         $title = $this->mLinkHolders['titles'][$key];
4236                                                         if ( is_null( $title ) ) continue;
4237
4238                                                         $pdbk = $title->getPrefixedDBkey();
4239
4240                                                         if(!isset($colours[$pdbk])){
4241                                                                 // found link in some of the variants, replace the link holder data
4242                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4243                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4244
4245                                                                 // set pdbk and colour
4246                                                                 $pdbks[$key] = $varPdbk;
4247                                                                 if ( $threshold >  0 ) {
4248                                                                         $size = $s->page_len;
4249                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4250                                                                                 $colours[$varPdbk] = 1;
4251                                                                         } else {
4252                                                                                 $colours[$varPdbk] = 2;
4253                                                                         }
4254                                                                 }
4255                                                                 else {
4256                                                                         $colours[$varPdbk] = 1;
4257                                                                 }
4258                                                         }
4259                                                 }
4260
4261                                                 // check if the object is a variant of a category
4262                                                 if(isset($categoryMap[$vardbk])){
4263                                                         $oldkey = $categoryMap[$vardbk];
4264                                                         if($oldkey != $vardbk)
4265                                                                 $varCategories[$oldkey]=$vardbk;
4266                                                 }
4267                                         }
4268
4269                                         // rebuild the categories in original order (if there are replacements)
4270                                         if(count($varCategories)>0){
4271                                                 $newCats = array();
4272                                                 $originalCats = $this->mOutput->getCategories();
4273                                                 foreach($originalCats as $cat => $sortkey){
4274                                                         // make the replacement
4275                                                         if( array_key_exists($cat,$varCategories) )
4276                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4277                                                         else $newCats[$cat] = $sortkey;
4278                                                 }
4279                                                 $this->mOutput->setCategoryLinks($newCats);
4280                                         }
4281                                 }
4282                         }
4283
4284                         # Construct search and replace arrays
4285                         wfProfileIn( $fname.'-construct' );
4286                         $replacePairs = array();
4287                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4288                                 $pdbk = $pdbks[$key];
4289                                 $searchkey = "<!--LINK $key-->";
4290                                 $title = $this->mLinkHolders['titles'][$key];
4291                                 if ( empty( $colours[$pdbk] ) ) {
4292                                         $linkCache->addBadLinkObj( $title );
4293                                         $colours[$pdbk] = 0;
4294                                         $this->mOutput->addLink( $title, 0 );
4295                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4296                                                                         $this->mLinkHolders['texts'][$key],
4297                                                                         $this->mLinkHolders['queries'][$key] );
4298                                 } elseif ( $colours[$pdbk] == 1 ) {
4299                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4300                                                                         $this->mLinkHolders['texts'][$key],
4301                                                                         $this->mLinkHolders['queries'][$key] );
4302                                 } elseif ( $colours[$pdbk] == 2 ) {
4303                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4304                                                                         $this->mLinkHolders['texts'][$key],
4305                                                                         $this->mLinkHolders['queries'][$key] );
4306                                 }
4307                         }
4308                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4309                         wfProfileOut( $fname.'-construct' );
4310
4311                         # Do the thing
4312                         wfProfileIn( $fname.'-replace' );
4313                         $text = preg_replace_callback(
4314                                 '/(<!--LINK .*?-->)/',
4315                                 $replacer->cb(),
4316                                 $text);
4317
4318                         wfProfileOut( $fname.'-replace' );
4319                 }
4320
4321                 # Now process interwiki link holders
4322                 # This is quite a bit simpler than internal links
4323                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4324                         wfProfileIn( $fname.'-interwiki' );
4325                         # Make interwiki link HTML
4326                         $replacePairs = array();
4327                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4328                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4329                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4330                         }
4331                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4332
4333                         $text = preg_replace_callback(
4334                                 '/<!--IWLINK (.*?)-->/',
4335                                 $replacer->cb(),
4336                                 $text );
4337                         wfProfileOut( $fname.'-interwiki' );
4338                 }
4339
4340                 wfProfileOut( $fname );
4341                 return $colours;
4342         }
4343
4344         /**
4345          * Replace <!--LINK--> link placeholders with plain text of links
4346          * (not HTML-formatted).
4347          * @param string $text
4348          * @return string
4349          */
4350         function replaceLinkHoldersText( $text ) {
4351                 $fname = 'Parser::replaceLinkHoldersText';
4352                 wfProfileIn( $fname );
4353
4354                 $text = preg_replace_callback(
4355                         '/<!--(LINK|IWLINK) (.*?)-->/',
4356                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4357                         $text );
4358
4359                 wfProfileOut( $fname );
4360                 return $text;
4361         }
4362
4363         /**
4364          * @param array $matches
4365          * @return string
4366          * @private
4367          */
4368         function replaceLinkHoldersTextCallback( $matches ) {
4369                 $type = $matches[1];
4370                 $key  = $matches[2];
4371                 if( $type == 'LINK' ) {
4372                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4373                                 return $this->mLinkHolders['texts'][$key];
4374                         }
4375                 } elseif( $type == 'IWLINK' ) {
4376                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4377                                 return $this->mInterwikiLinkHolders['texts'][$key];
4378                         }
4379                 }
4380                 return $matches[0];
4381         }
4382
4383         /**
4384          * Tag hook handler for 'pre'.
4385          */
4386         function renderPreTag( $text, $attribs ) {
4387                 // Backwards-compatibility hack
4388                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4389
4390                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4391                 return wfOpenElement( 'pre', $attribs ) .
4392                         Xml::escapeTagsOnly( $content ) .
4393                         '</pre>';
4394         }
4395
4396         /**
4397          * Renders an image gallery from a text with one line per image.
4398          * text labels may be given by using |-style alternative text. E.g.
4399          *   Image:one.jpg|The number "1"
4400          *   Image:tree.jpg|A tree
4401          * given as text will return the HTML of a gallery with two images,
4402          * labeled 'The number "1"' and
4403          * 'A tree'.
4404          */
4405         function renderImageGallery( $text, $params ) {
4406                 $ig = new ImageGallery();
4407                 $ig->setContextTitle( $this->mTitle );
4408                 $ig->setShowBytes( false );
4409                 $ig->setShowFilename( false );
4410                 $ig->setParser( $this );
4411                 $ig->setHideBadImages();
4412                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4413                 $ig->useSkin( $this->mOptions->getSkin() );
4414                 $ig->mRevisionId = $this->mRevisionId;
4415
4416                 if( isset( $params['caption'] ) ) {
4417                         $caption = $params['caption'];
4418                         $caption = htmlspecialchars( $caption );
4419                         $caption = $this->replaceInternalLinks( $caption );
4420                         $ig->setCaptionHtml( $caption );
4421                 }
4422                 if( isset( $params['perrow'] ) ) {
4423                         $ig->setPerRow( $params['perrow'] );
4424                 }
4425                 if( isset( $params['widths'] ) ) {
4426                         $ig->setWidths( $params['widths'] );
4427                 }
4428                 if( isset( $params['heights'] ) ) {
4429                         $ig->setHeights( $params['heights'] );
4430                 }
4431
4432                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4433
4434                 $lines = explode( "\n", $text );
4435                 foreach ( $lines as $line ) {
4436                         # match lines like these:
4437                         # Image:someimage.jpg|This is some image
4438                         $matches = array();
4439                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4440                         # Skip empty lines
4441                         if ( count( $matches ) == 0 ) {
4442                                 continue;
4443                         }
4444                         $tp = Title::newFromText( $matches[1] );
4445                         $nt =& $tp;
4446                         if( is_null( $nt ) ) {
4447                                 # Bogus title. Ignore these so we don't bomb out later.
4448                                 continue;
4449                         }
4450                         if ( isset( $matches[3] ) ) {
4451                                 $label = $matches[3];
4452                         } else {
4453                                 $label = '';
4454                         }
4455
4456                         $pout = $this->parse( $label,
4457                                 $this->mTitle,
4458                                 $this->mOptions,
4459                                 false, // Strip whitespace...?
4460                                 false  // Don't clear state!
4461                         );
4462                         $html = $pout->getText();
4463
4464                         $ig->add( $nt, $html );
4465
4466                         # Only add real images (bug #5586)
4467                         if ( $nt->getNamespace() == NS_IMAGE ) {
4468                                 $this->mOutput->addImage( $nt->getDBkey() );
4469                         }
4470                 }
4471                 return $ig->toHTML();
4472         }
4473
4474         function getImageParams( $handler ) {
4475                 if ( $handler ) {
4476                         $handlerClass = get_class( $handler );
4477                 } else {
4478                         $handlerClass = '';
4479                 }
4480                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4481                         // Initialise static lists
4482                         static $internalParamNames = array(
4483                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4484                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4485                                         'bottom', 'text-bottom' ),
4486                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4487                                         'upright', 'border' ),
4488                         );
4489                         static $internalParamMap;
4490                         if ( !$internalParamMap ) {
4491                                 $internalParamMap = array();
4492                                 foreach ( $internalParamNames as $type => $names ) {
4493                                         foreach ( $names as $name ) {
4494                                                 $magicName = str_replace( '-', '_', "img_$name" );
4495                                                 $internalParamMap[$magicName] = array( $type, $name );
4496                                         }
4497                                 }
4498                         }
4499
4500                         // Add handler params
4501                         $paramMap = $internalParamMap;
4502                         if ( $handler ) {
4503                                 $handlerParamMap = $handler->getParamMap();
4504                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4505                                         $paramMap[$magic] = array( 'handler', $paramName );
4506                                 }
4507                         }
4508                         $this->mImageParams[$handlerClass] = $paramMap;
4509                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4510                 }
4511                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4512         }
4513
4514         /**
4515          * Parse image options text and use it to make an image
4516          */
4517         function makeImage( $title, $options ) {
4518                 # @TODO: let the MediaHandler specify its transform parameters
4519                 #
4520                 # Check if the options text is of the form "options|alt text"
4521                 # Options are:
4522                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4523                 #  * left               no resizing, just left align. label is used for alt= only
4524                 #  * right              same, but right aligned
4525                 #  * none               same, but not aligned
4526                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4527                 #  * center             center the image
4528                 #  * framed             Keep original image size, no magnify-button.
4529                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4530                 #  * upright            reduce width for upright images, rounded to full __0 px
4531                 #  * border             draw a 1px border around the image
4532                 # vertical-align values (no % or length right now):
4533                 #  * baseline
4534                 #  * sub
4535                 #  * super
4536                 #  * top
4537                 #  * text-top
4538                 #  * middle
4539                 #  * bottom
4540                 #  * text-bottom
4541
4542                 $parts = array_map( 'trim', explode( '|', $options) );
4543                 $sk = $this->mOptions->getSkin();
4544
4545                 # Give extensions a chance to select the file revision for us
4546                 $skip = $time = false;
4547                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4548
4549                 if ( $skip ) {
4550                         return $sk->makeLinkObj( $title );
4551                 }
4552
4553                 # Get parameter map
4554                 $file = wfFindFile( $title, $time );
4555                 $handler = $file ? $file->getHandler() : false;
4556
4557                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4558
4559                 # Process the input parameters
4560                 $caption = '';
4561                 $params = array( 'frame' => array(), 'handler' => array(),
4562                         'horizAlign' => array(), 'vertAlign' => array() );
4563                 foreach( $parts as $part ) {
4564                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4565                         if ( isset( $paramMap[$magicName] ) ) {
4566                                 list( $type, $paramName ) = $paramMap[$magicName];
4567                                 $params[$type][$paramName] = $value;
4568
4569                                 // Special case; width and height come in one variable together
4570                                 if( $type == 'handler' && $paramName == 'width' ) {
4571                                         $m = array();
4572                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4573                                                 $params[$type]['width'] = intval( $m[1] );
4574                                                 $params[$type]['height'] = intval( $m[2] );
4575                                         } else {
4576                                                 $params[$type]['width'] = intval( $value );
4577                                         }
4578                                 }
4579                         } else {
4580                                 $caption = $part;
4581                         }
4582                 }
4583
4584                 # Process alignment parameters
4585                 if ( $params['horizAlign'] ) {
4586                         $params['frame']['align'] = key( $params['horizAlign'] );
4587                 }
4588                 if ( $params['vertAlign'] ) {
4589                         $params['frame']['valign'] = key( $params['vertAlign'] );
4590                 }
4591
4592                 # Validate the handler parameters
4593                 if ( $handler ) {
4594                         foreach ( $params['handler'] as $name => $value ) {
4595                                 if ( !$handler->validateParam( $name, $value ) ) {
4596                                         unset( $params['handler'][$name] );
4597                                 }
4598                         }
4599                 }
4600
4601                 # Strip bad stuff out of the alt text
4602                 $alt = $this->replaceLinkHoldersText( $caption );
4603
4604                 # make sure there are no placeholders in thumbnail attributes
4605                 # that are later expanded to html- so expand them now and
4606                 # remove the tags
4607                 $alt = $this->mStripState->unstripBoth( $alt );
4608                 $alt = Sanitizer::stripAllTags( $alt );
4609
4610                 $params['frame']['alt'] = $alt;
4611                 $params['frame']['caption'] = $caption;
4612
4613                 # Linker does the rest
4614                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4615
4616                 # Give the handler a chance to modify the parser object
4617                 if ( $handler ) {
4618                         $handler->parserTransformHook( $this, $file );
4619                 }
4620
4621                 return $ret;
4622         }
4623
4624         /**
4625          * Set a flag in the output object indicating that the content is dynamic and
4626          * shouldn't be cached.
4627          */
4628         function disableCache() {
4629                 wfDebug( "Parser output marked as uncacheable.\n" );
4630                 $this->mOutput->mCacheTime = -1;
4631         }
4632
4633         /**#@+
4634          * Callback from the Sanitizer for expanding items found in HTML attribute
4635          * values, so they can be safely tested and escaped.
4636          * @param string $text
4637          * @param array $args
4638          * @return string
4639          * @private
4640          */
4641         function attributeStripCallback( &$text, $args ) {
4642                 $text = $this->replaceVariables( $text, $args );
4643                 $text = $this->mStripState->unstripBoth( $text );
4644                 return $text;
4645         }
4646
4647         /**#@-*/
4648
4649         /**#@+
4650          * Accessor/mutator
4651          */
4652         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4653         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4654         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4655         /**#@-*/
4656
4657         /**#@+
4658          * Accessor
4659          */
4660         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4661         /**#@-*/
4662
4663
4664         /**
4665          * Break wikitext input into sections, and either pull or replace
4666          * some particular section's text.
4667          *
4668          * External callers should use the getSection and replaceSection methods.
4669          *
4670          * @param $text Page wikitext
4671          * @param $section Numbered section. 0 pulls the text before the first
4672          *                 heading; other numbers will pull the given section
4673          *                 along with its lower-level subsections.
4674          * @param $mode One of "get" or "replace"
4675          * @param $newtext Replacement text for section data.
4676          * @return string for "get", the extracted section text.
4677          *                for "replace", the whole page with the section replaced.
4678          */
4679         private function extractSections( $text, $section, $mode, $newtext='' ) {
4680                 # I.... _hope_ this is right.
4681                 # Otherwise, sometimes we don't have things initialized properly.
4682                 $this->clearState();
4683
4684                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4685                 # comments to be stripped as well)
4686                 $stripState = new StripState;
4687
4688                 $oldOutputType = $this->mOutputType;
4689                 $oldOptions = $this->mOptions;
4690                 $this->mOptions = new ParserOptions();
4691                 $this->setOutputType( OT_WIKI );
4692
4693                 $striptext = $this->strip( $text, $stripState, true );
4694
4695                 $this->setOutputType( $oldOutputType );
4696                 $this->mOptions = $oldOptions;
4697
4698                 # now that we can be sure that no pseudo-sections are in the source,
4699                 # split it up by section
4700                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4701                 $comment = "(?:$uniq-!--.*?QINU\x07)";
4702                 $secs = preg_split(
4703                         "/
4704                         (
4705                                 ^
4706                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4707                                 (=+) # Should this be limited to 6?
4708                                 .+?  # Section title...
4709                                 \\2  # Ending = count must match start
4710                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4711                                 $
4712                         |
4713                                 <h([1-6])\b.*?>
4714                                 .*?
4715                                 <\/h\\3\s*>
4716                         )
4717                         /mix",
4718                         $striptext, -1,
4719                         PREG_SPLIT_DELIM_CAPTURE);
4720
4721                 if( $mode == "get" ) {
4722                         if( $section == 0 ) {
4723                                 // "Section 0" returns the content before any other section.
4724                                 $rv = $secs[0];
4725                         } else {
4726                                 //track missing section, will replace if found.
4727                                 $rv = $newtext;
4728                         }
4729                 } elseif( $mode == "replace" ) {
4730                         if( $section == 0 ) {
4731                                 $rv = $newtext . "\n\n";
4732                                 $remainder = true;
4733                         } else {
4734                                 $rv = $secs[0];
4735                                 $remainder = false;
4736                         }
4737                 }
4738                 $count = 0;
4739                 $sectionLevel = 0;
4740                 for( $index = 1; $index < count( $secs ); ) {
4741                         $headerLine = $secs[$index++];
4742                         if( $secs[$index] ) {
4743                                 // A wiki header
4744                                 $headerLevel = strlen( $secs[$index++] );
4745                         } else {
4746                                 // An HTML header
4747                                 $index++;
4748                                 $headerLevel = intval( $secs[$index++] );
4749                         }
4750                         $content = $secs[$index++];
4751
4752                         $count++;
4753                         if( $mode == "get" ) {
4754                                 if( $count == $section ) {
4755                                         $rv = $headerLine . $content;
4756                                         $sectionLevel = $headerLevel;
4757                                 } elseif( $count > $section ) {
4758                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4759                                                 $rv .= $headerLine . $content;
4760                                         } else {
4761                                                 // Broke out to a higher-level section
4762                                                 break;
4763                                         }
4764                                 }
4765                         } elseif( $mode == "replace" ) {
4766                                 if( $count < $section ) {
4767                                         $rv .= $headerLine . $content;
4768                                 } elseif( $count == $section ) {
4769                                         $rv .= $newtext . "\n\n";
4770                                         $sectionLevel = $headerLevel;
4771                                 } elseif( $count > $section ) {
4772                                         if( $headerLevel <= $sectionLevel ) {
4773                                                 // Passed the section's sub-parts.
4774                                                 $remainder = true;
4775                                         }
4776                                         if( $remainder ) {
4777                                                 $rv .= $headerLine . $content;
4778                                         }
4779                                 }
4780                         }
4781                 }
4782                 if (is_string($rv))
4783                         # reinsert stripped tags
4784                         $rv = trim( $stripState->unstripBoth( $rv ) );
4785
4786                 return $rv;
4787         }
4788
4789         /**
4790          * This function returns the text of a section, specified by a number ($section).
4791          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4792          * the first section before any such heading (section 0).
4793          *
4794          * If a section contains subsections, these are also returned.
4795          *
4796          * @param $text String: text to look in
4797          * @param $section Integer: section number
4798          * @param $deftext: default to return if section is not found
4799          * @return string text of the requested section
4800          */
4801         public function getSection( $text, $section, $deftext='' ) {
4802                 return $this->extractSections( $text, $section, "get", $deftext );
4803         }
4804
4805         public function replaceSection( $oldtext, $section, $text ) {
4806                 return $this->extractSections( $oldtext, $section, "replace", $text );
4807         }
4808
4809         /**
4810          * Get the timestamp associated with the current revision, adjusted for
4811          * the default server-local timestamp
4812          */
4813         function getRevisionTimestamp() {
4814                 if ( is_null( $this->mRevisionTimestamp ) ) {
4815                         wfProfileIn( __METHOD__ );
4816                         global $wgContLang;
4817                         $dbr = wfGetDB( DB_SLAVE );
4818                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4819                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4820
4821                         // Normalize timestamp to internal MW format for timezone processing.
4822                         // This has the added side-effect of replacing a null value with
4823                         // the current time, which gives us more sensible behavior for
4824                         // previews.
4825                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4826
4827                         // The cryptic '' timezone parameter tells to use the site-default
4828                         // timezone offset instead of the user settings.
4829                         //
4830                         // Since this value will be saved into the parser cache, served
4831                         // to other users, and potentially even used inside links and such,
4832                         // it needs to be consistent for all visitors.
4833                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4834
4835                         wfProfileOut( __METHOD__ );
4836                 }
4837                 return $this->mRevisionTimestamp;
4838         }
4839
4840         /**
4841          * Mutator for $mDefaultSort
4842          *
4843          * @param $sort New value
4844          */
4845         public function setDefaultSort( $sort ) {
4846                 $this->mDefaultSort = $sort;
4847         }
4848
4849         /**
4850          * Accessor for $mDefaultSort
4851          * Will use the title/prefixed title if none is set
4852          *
4853          * @return string
4854          */
4855         public function getDefaultSort() {
4856                 if( $this->mDefaultSort !== false ) {
4857                         return $this->mDefaultSort;
4858                 } else {
4859                         return $this->mTitle->getNamespace() == NS_CATEGORY
4860                                         ? $this->mTitle->getText()
4861                                         : $this->mTitle->getPrefixedText();
4862                 }
4863         }
4864
4865         /**
4866          * Try to guess the section anchor name based on a wikitext fragment
4867          * presumably extracted from a heading, for example "Header" from
4868          * "== Header ==".
4869          */
4870         public function guessSectionNameFromWikiText( $text ) {
4871                 # Strip out wikitext links(they break the anchor)
4872                 $text = $this->stripSectionName( $text );
4873                 $headline = Sanitizer::decodeCharReferences( $text );
4874                 # strip out HTML
4875                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4876                 $headline = trim( $headline );
4877                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4878                 $replacearray = array(
4879                         '%3A' => ':',
4880                         '%' => '.'
4881                 );
4882                 return str_replace(
4883                         array_keys( $replacearray ),
4884                         array_values( $replacearray ),
4885                         $sectionanchor );
4886         }
4887
4888         /**
4889          * Strips a text string of wikitext for use in a section anchor
4890          *
4891          * Accepts a text string and then removes all wikitext from the
4892          * string and leaves only the resultant text (i.e. the result of
4893          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4894          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4895          * to create valid section anchors by mimicing the output of the
4896          * parser when headings are parsed.
4897          *
4898          * @param $text string Text string to be stripped of wikitext
4899          * for use in a Section anchor
4900          * @return Filtered text string
4901          */
4902         public function stripSectionName( $text ) {
4903                 # Strip internal link markup
4904                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4905                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4906
4907                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4908                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4909                 # on how many empty links there are on the page - need to figure that out.
4910                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4911
4912                 # Parse wikitext quotes (italics & bold)
4913                 $text = $this->doQuotes($text);
4914
4915                 # Strip HTML tags
4916                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4917                 return $text;
4918         }
4919
4920         /**
4921          * strip/replaceVariables/unstrip for preprocessor regression testing
4922          */
4923         function srvus( $text ) {
4924                 $text = $this->strip( $text, $this->mStripState );
4925                 $text = Sanitizer::removeHTMLtags( $text );
4926                 $text = $this->replaceVariables( $text );
4927                 $text = preg_replace( '/<!--MWTEMPLATESECTION.*?-->/', '', $text );
4928                 $text = $this->mStripState->unstripBoth( $text );
4929                 return $text;
4930         }
4931 }
4932