includes/Parser_OldPP.php

   1 <?php
   2 /**
   3  * Parser with old preprocessor
   4  * @ingroup Parser
   5  */
   6 class Parser_OldPP
   7 {
   8         /**
   9          * Update this version number when the ParserOutput format
  10          * changes in an incompatible way, so the parser cache
  11          * can automatically discard old data.
  12          */
  13         const VERSION = '1.6.4';
  14
  15         # Flags for Parser::setFunctionHook
  16         # Also available as global constants from Defines.php
  17         const SFH_NO_HASH = 1;
  18
  19         # Constants needed for external link processing
  20         # Everything except bracket, space, or control characters
  21         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  22         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)\\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/S';
  23
  24         // State constants for the definition list colon extraction
  25         const COLON_STATE_TEXT = 0;
  26         const COLON_STATE_TAG = 1;
  27         const COLON_STATE_TAGSTART = 2;
  28         const COLON_STATE_CLOSETAG = 3;
  29         const COLON_STATE_TAGSLASH = 4;
  30         const COLON_STATE_COMMENT = 5;
  31         const COLON_STATE_COMMENTDASH = 6;
  32         const COLON_STATE_COMMENTDASHDASH = 7;
  33
  34         // Allowed values for $this->mOutputType
  35         // Parameter to startExternalParse().
  36         const OT_HTML = 1;
  37         const OT_WIKI = 2;
  38         const OT_PREPROCESS = 3;
  39         const OT_MSG = 4;
  40
  41         /**#@+
  42          * @private
  43          */
  44         # Persistent:
  45         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  46                 $mImageParams, $mImageParamsMagicArray, $mExtLinkBracketedRegex;
  47
  48         # Cleared with clearState():
  49         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  50         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  51         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
  52         var $mIncludeSizes, $mDefaultSort;
  53         var $mTemplates,        // cache of already loaded templates, avoids
  54                                 // multiple SQL queries for the same string
  55             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  56                                 // in this path. Used for loop detection.
  57
  58         # Temporary
  59         # These are variables reset at least once per parse regardless of $clearState
  60         var $mOptions,      // ParserOptions object
  61                 $mTitle,        // Title context, used for self-link rendering and similar things
  62                 $mOutputType,   // Output type, one of the OT_xxx constants
  63                 $ot,            // Shortcut alias, see setOutputType()
  64                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  65                 $mRevisionTimestamp, // The timestamp of the specified revision ID
  66                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
  67
  68         /**#@-*/
  69
  70         /**
  71          * Constructor
  72          *
  73          * @public
  74          */
  75         function __construct( $conf = array() ) {
  76                 $this->mTagHooks = array();
  77                 $this->mTransparentTagHooks = array();
  78                 $this->mFunctionHooks = array();
  79                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
  80                 $this->mFirstCall = true;
  81                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
  82                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
  83         }
  84
  85         /**
  86          * Do various kinds of initialisation on the first call of the parser
  87          */
  88         function firstCallInit() {
  89                 if ( !$this->mFirstCall ) {
  90                         return;
  91                 }
  92                 $this->mFirstCall = false;
  93
  94                 wfProfileIn( __METHOD__ );
  95                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
  96
  97                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
  98
  99                 # Syntax for arguments (see self::setFunctionHook):
 100                 #  "name for lookup in localized magic words array",
 101                 #  function callback,
 102                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 103                 #    instead of {{#int:...}})
 104                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 105                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 106                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 107                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 108                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 109                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 110                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 111                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 112                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 113                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 114                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 115                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 116                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 117                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 118                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 119                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 120                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 121                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 122                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 123                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 124                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 125                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 126                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 127                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 128                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 129                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 130                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 131
 132                 if ( $wgAllowDisplayTitle ) {
 133                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 134                 }
 135                 if ( $wgAllowSlowParserFunctions ) {
 136                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 137                 }
 138
 139                 $this->initialiseVariables();
 140
 141                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 142                 wfProfileOut( __METHOD__ );
 143         }
 144
 145         /**
 146          * Clear Parser state
 147          *
 148          * @private
 149          */
 150         function clearState() {
 151                 wfProfileIn( __METHOD__ );
 152                 if ( $this->mFirstCall ) {
 153                         $this->firstCallInit();
 154                 }
 155                 $this->mOutput = new ParserOutput;
 156                 $this->mAutonumber = 0;
 157                 $this->mLastSection = '';
 158                 $this->mDTopen = false;
 159                 $this->mIncludeCount = array();
 160                 $this->mStripState = new StripState;
 161                 $this->mArgStack = array();
 162                 $this->mInPre = false;
 163                 $this->mInterwikiLinkHolders = array(
 164                         'texts' => array(),
 165                         'titles' => array()
 166                 );
 167                 $this->mLinkHolders = array(
 168                         'namespaces' => array(),
 169                         'dbkeys' => array(),
 170                         'queries' => array(),
 171                         'texts' => array(),
 172                         'titles' => array()
 173                 );
 174                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 175
 176                 /**
 177                  * Prefix for temporary replacement strings for the multipass parser.
 178                  * \x07 should never appear in input as it's disallowed in XML.
 179                  * Using it at the front also gives us a little extra robustness
 180                  * since it shouldn't match when butted up against identifier-like
 181                  * string constructs.
 182                  */
 183                 $this->mUniqPrefix = "\x07UNIQ" . self::getRandomString();
 184
 185                 # Clear these on every parse, bug 4549
 186                 $this->mTemplates = array();
 187                 $this->mTemplatePath = array();
 188
 189                 $this->mShowToc = true;
 190                 $this->mForceTocPosition = false;
 191                 $this->mIncludeSizes = array(
 192                         'pre-expand' => 0,
 193                         'post-expand' => 0,
 194                         'arg' => 0
 195                 );
 196                 $this->mDefaultSort = false;
 197
 198                 wfRunHooks( 'ParserClearState', array( &$this ) );
 199                 wfProfileOut( __METHOD__ );
 200         }
 201
 202         function setOutputType( $ot ) {
 203                 $this->mOutputType = $ot;
 204                 // Shortcut alias
 205                 $this->ot = array(
 206                         'html' => $ot == self::OT_HTML,
 207                         'wiki' => $ot == self::OT_WIKI,
 208                         'msg' => $ot == self::OT_MSG,
 209                         'pre' => $ot == self::OT_PREPROCESS,
 210                 );
 211         }
 212
 213         /**
 214          * Accessor for mUniqPrefix.
 215          *
 216          * @public
 217          */
 218         function uniqPrefix() {
 219                 return $this->mUniqPrefix;
 220         }
 221
 222         /**
 223          * Convert wikitext to HTML
 224          * Do not call this function recursively.
 225          *
 226          * @param string $text Text we want to parse
 227          * @param Title &$title A title object
 228          * @param array $options
 229          * @param boolean $linestart
 230          * @param boolean $clearState
 231          * @param int $revid number to pass in {{REVISIONID}}
 232          * @return ParserOutput a ParserOutput
 233          */
 234         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 235                 /**
 236                  * First pass--just handle <nowiki> sections, pass the rest off
 237                  * to internalParse() which does all the real work.
 238                  */
 239
 240                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 241                 $fname = 'Parser::parse-' . wfGetCaller();
 242                 wfProfileIn( __METHOD__ );
 243                 wfProfileIn( $fname );
 244
 245                 if ( $clearState ) {
 246                         $this->clearState();
 247                 }
 248
 249                 $this->mOptions = $options;
 250                 $this->mTitle =& $title;
 251                 $oldRevisionId = $this->mRevisionId;
 252                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 253                 if( $revid !== null ) {
 254                         $this->mRevisionId = $revid;
 255                         $this->mRevisionTimestamp = null;
 256                 }
 257                 $this->setOutputType( self::OT_HTML );
 258                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 259                 $text = $this->strip( $text, $this->mStripState );
 260                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 261                 $text = $this->internalParse( $text );
 262                 $text = $this->mStripState->unstripGeneral( $text );
 263
 264                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 265                 $fixtags = array(
 266                         # french spaces, last one Guillemet-left
 267                         # only if there is something before the space
 268                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 269                         # french spaces, Guillemet-right
 270                         '/(\\302\\253) /' => '\\1&nbsp;',
 271                 );
 272                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 273
 274                 # only once and last
 275                 $text = $this->doBlockLevels( $text, $linestart );
 276
 277                 $this->replaceLinkHolders( $text );
 278
 279                 # the position of the parserConvert() call should not be changed. it
 280                 # assumes that the links are all replaced and the only thing left
 281                 # is the <nowiki> mark.
 282                 # Side-effects: this calls $this->mOutput->setTitleText()
 283                 $text = $wgContLang->parserConvert( $text, $this );
 284
 285                 $text = $this->mStripState->unstripNoWiki( $text );
 286
 287                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 288
 289 //!JF Move to its own function
 290
 291                 $uniq_prefix = $this->mUniqPrefix;
 292                 $matches = array();
 293                 $elements = array_keys( $this->mTransparentTagHooks );
 294                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 295
 296                 foreach( $matches as $marker => $data ) {
 297                         list( $element, $content, $params, $tag ) = $data;
 298                         $tagName = strtolower( $element );
 299                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 300                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 301                                         array( $content, $params, $this ) );
 302                         } else {
 303                                 $output = $tag;
 304                         }
 305                         $this->mStripState->general->setPair( $marker, $output );
 306                 }
 307                 $text = $this->mStripState->unstripGeneral( $text );
 308
 309                 $text = Sanitizer::normalizeCharReferences( $text );
 310
 311                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 312                         $text = self::tidy($text);
 313                 } else {
 314                         # attempt to sanitize at least some nesting problems
 315                         # (bug #2702 and quite a few others)
 316                         $tidyregs = array(
 317                                 # ''Something [http://www.cool.com cool''] -->
 318                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 319                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 320                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 321                                 # fix up an anchor inside another anchor, only
 322                                 # at least for a single single nested link (bug 3695)
 323                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 324                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 325                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 326                                 # contains a div, so fix it up here; replace
 327                                 # div with escaped text
 328                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 329                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 330                                 # remove empty italic or bold tag pairs, some
 331                                 # introduced by rules above
 332                                 '/<([bi])><\/\\1>/' => '',
 333                         );
 334
 335                         $text = preg_replace(
 336                                 array_keys( $tidyregs ),
 337                                 array_values( $tidyregs ),
 338                                 $text );
 339                 }
 340
 341                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 342
 343                 # Information on include size limits, for the benefit of users who try to skirt them
 344                 if ( $this->mOptions->getEnableLimitReport() ) {
 345                         $max = $this->mOptions->getMaxIncludeSize();
 346                         $limitReport =
 347                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']}/$max bytes\n" .
 348                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 349                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 350                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 351                         $text .= "<!-- \n$limitReport-->\n";
 352                 }
 353                 $this->mOutput->setText( $text );
 354                 $this->mRevisionId = $oldRevisionId;
 355                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 356                 wfProfileOut( $fname );
 357                 wfProfileOut( __METHOD__ );
 358
 359                 return $this->mOutput;
 360         }
 361
 362         /**
 363          * Recursive parser entry point that can be called from an extension tag
 364          * hook.
 365          */
 366         function recursiveTagParse( $text ) {
 367                 wfProfileIn( __METHOD__ );
 368                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 369                 $text = $this->strip( $text, $this->mStripState );
 370                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 371                 $text = $this->internalParse( $text );
 372                 wfProfileOut( __METHOD__ );
 373                 return $text;
 374         }
 375
 376         /**
 377          * Expand templates and variables in the text, producing valid, static wikitext.
 378          * Also removes comments.
 379          */
 380         function preprocess( $text, $title, $options, $revid = null ) {
 381                 wfProfileIn( __METHOD__ );
 382                 $this->clearState();
 383                 $this->setOutputType( self::OT_PREPROCESS );
 384                 $this->mOptions = $options;
 385                 $this->mTitle = $title;
 386                 if( $revid !== null ) {
 387                         $this->mRevisionId = $revid;
 388                 }
 389                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 390                 $text = $this->strip( $text, $this->mStripState );
 391                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 392                 if ( $this->mOptions->getRemoveComments() ) {
 393                         $text = Sanitizer::removeHTMLcomments( $text );
 394                 }
 395                 $text = $this->replaceVariables( $text );
 396                 $text = $this->mStripState->unstripBoth( $text );
 397                 wfProfileOut( __METHOD__ );
 398                 return $text;
 399         }
 400
 401         /**
 402          * Get a random string
 403          *
 404          * @private
 405          * @static
 406          */
 407         function getRandomString() {
 408                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 409         }
 410
 411         function &getTitle() { return $this->mTitle; }
 412         function getOptions() { return $this->mOptions; }
 413         function getRevisionId() { return $this->mRevisionId; }
 414
 415         function getFunctionLang() {
 416                 global $wgLang, $wgContLang;
 417                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 418         }
 419
 420         /**
 421          * Replaces all occurrences of HTML-style comments and the given tags
 422          * in the text with a random marker and returns teh next text. The output
 423          * parameter $matches will be an associative array filled with data in
 424          * the form:
 425          *   'UNIQ-xxxxx' => array(
 426          *     'element',
 427          *     'tag content',
 428          *     array( 'param' => 'x' ),
 429          *     '<element param="x">tag content</element>' ) )
 430          *
 431          * @param $elements list of element names. Comments are always extracted.
 432          * @param $text Source text string.
 433          * @param $uniq_prefix
 434          *
 435          * @public
 436          * @static
 437          */
 438         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 439                 static $n = 1;
 440                 $stripped = '';
 441                 $matches = array();
 442
 443                 $taglist = implode( '|', $elements );
 444                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 445
 446                 while ( '' != $text ) {
 447                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 448                         $stripped .= $p[0];
 449                         if( count( $p ) < 5 ) {
 450                                 break;
 451                         }
 452                         if( count( $p ) > 5 ) {
 453                                 // comment
 454                                 $element    = $p[4];
 455                                 $attributes = '';
 456                                 $close      = '';
 457                                 $inside     = $p[5];
 458                         } else {
 459                                 // tag
 460                                 $element    = $p[1];
 461                                 $attributes = $p[2];
 462                                 $close      = $p[3];
 463                                 $inside     = $p[4];
 464                         }
 465
 466                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . "-QINU\x07";
 467                         $stripped .= $marker;
 468
 469                         if ( $close === '/>' ) {
 470                                 // Empty element tag, <tag />
 471                                 $content = null;
 472                                 $text = $inside;
 473                                 $tail = null;
 474                         } else {
 475                                 if( $element == '!--' ) {
 476                                         $end = '/(-->)/';
 477                                 } else {
 478                                         $end = "/(<\\/$element\\s*>)/i";
 479                                 }
 480                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 481                                 $content = $q[0];
 482                                 if( count( $q ) < 3 ) {
 483                                         # No end tag -- let it run out to the end of the text.
 484                                         $tail = '';
 485                                         $text = '';
 486                                 } else {
 487                                         $tail = $q[1];
 488                                         $text = $q[2];
 489                                 }
 490                         }
 491
 492                         $matches[$marker] = array( $element,
 493                                 $content,
 494                                 Sanitizer::decodeTagAttributes( $attributes ),
 495                                 "<$element$attributes$close$content$tail" );
 496                 }
 497                 return $stripped;
 498         }
 499
 500         /**
 501          * Strips and renders nowiki, pre, math, hiero
 502          * If $render is set, performs necessary rendering operations on plugins
 503          * Returns the text, and fills an array with data needed in unstrip()
 504          *
 505          * @param StripState $state
 506          *
 507          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 508          *  will be stripped in addition to other tags. This is important
 509          *  for section editing, where these comments cause confusion when
 510          *  counting the sections in the wikisource
 511          *
 512          * @param array dontstrip contains tags which should not be stripped;
 513          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 514          *
 515          * @private
 516          */
 517         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 518                 global $wgContLang;
 519                 wfProfileIn( __METHOD__ );
 520                 $render = ($this->mOutputType == self::OT_HTML);
 521
 522                 $uniq_prefix = $this->mUniqPrefix;
 523                 $commentState = new ReplacementArray;
 524                 $nowikiItems = array();
 525                 $generalItems = array();
 526
 527                 $elements = array_merge(
 528                         array( 'nowiki', 'gallery' ),
 529                         array_keys( $this->mTagHooks ) );
 530                 global $wgRawHtml;
 531                 if( $wgRawHtml ) {
 532                         $elements[] = 'html';
 533                 }
 534                 if( $this->mOptions->getUseTeX() ) {
 535                         $elements[] = 'math';
 536                 }
 537
 538                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 539                 foreach ( $elements AS $k => $v ) {
 540                         if ( !in_array ( $v , $dontstrip ) ) continue;
 541                         unset ( $elements[$k] );
 542                 }
 543
 544                 $matches = array();
 545                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 546
 547                 foreach( $matches as $marker => $data ) {
 548                         list( $element, $content, $params, $tag ) = $data;
 549                         if( $render ) {
 550                                 $tagName = strtolower( $element );
 551                                 wfProfileIn( __METHOD__."-render-$tagName" );
 552                                 switch( $tagName ) {
 553                                 case '!--':
 554                                         // Comment
 555                                         if( substr( $tag, -3 ) == '-->' ) {
 556                                                 $output = $tag;
 557                                         } else {
 558                                                 // Unclosed comment in input.
 559                                                 // Close it so later stripping can remove it
 560                                                 $output = "$tag-->";
 561                                         }
 562                                         break;
 563                                 case 'html':
 564                                         if( $wgRawHtml ) {
 565                                                 $output = $content;
 566                                                 break;
 567                                         }
 568                                         // Shouldn't happen otherwise. :)
 569                                 case 'nowiki':
 570                                         $output = Xml::escapeTagsOnly( $content );
 571                                         break;
 572                                 case 'math':
 573                                         $output = $wgContLang->armourMath(
 574                                                 MathRenderer::renderMath( $content, $params ) );
 575                                         break;
 576                                 case 'gallery':
 577                                         $output = $this->renderImageGallery( $content, $params );
 578                                         break;
 579                                 default:
 580                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 581                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 582                                                         array( $content, $params, $this ) );
 583                                         } else {
 584                                                 throw new MWException( "Invalid call hook $element" );
 585                                         }
 586                                 }
 587                                 wfProfileOut( __METHOD__."-render-$tagName" );
 588                         } else {
 589                                 // Just stripping tags; keep the source
 590                                 $output = $tag;
 591                         }
 592
 593                         // Unstrip the output, to support recursive strip() calls
 594                         $output = $state->unstripBoth( $output );
 595
 596                         if( !$stripcomments && $element == '!--' ) {
 597                                 $commentState->setPair( $marker, $output );
 598                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 599                                 $nowikiItems[$marker] = $output;
 600                         } else {
 601                                 $generalItems[$marker] = $output;
 602                         }
 603                 }
 604                 # Add the new items to the state
 605                 # We do this after the loop instead of during it to avoid slowing
 606                 # down the recursive unstrip
 607                 $state->nowiki->mergeArray( $nowikiItems );
 608                 $state->general->mergeArray( $generalItems );
 609
 610                 # Unstrip comments unless explicitly told otherwise.
 611                 # (The comments are always stripped prior to this point, so as to
 612                 # not invoke any extension tags / parser hooks contained within
 613                 # a comment.)
 614                 if ( !$stripcomments ) {
 615                         // Put them all back and forget them
 616                         $text = $commentState->replace( $text );
 617                 }
 618
 619                 wfProfileOut( __METHOD__ );
 620                 return $text;
 621         }
 622
 623         /**
 624          * Restores pre, math, and other extensions removed by strip()
 625          *
 626          * always call unstripNoWiki() after this one
 627          * @private
 628          * @deprecated use $this->mStripState->unstrip()
 629          */
 630         function unstrip( $text, $state ) {
 631                 return $state->unstripGeneral( $text );
 632         }
 633
 634         /**
 635          * Always call this after unstrip() to preserve the order
 636          *
 637          * @private
 638          * @deprecated use $this->mStripState->unstrip()
 639          */
 640         function unstripNoWiki( $text, $state ) {
 641                 return $state->unstripNoWiki( $text );
 642         }
 643
 644         /**
 645          * @deprecated use $this->mStripState->unstripBoth()
 646          */
 647         function unstripForHTML( $text ) {
 648                 return $this->mStripState->unstripBoth( $text );
 649         }
 650
 651         /**
 652          * Add an item to the strip state
 653          * Returns the unique tag which must be inserted into the stripped text
 654          * The tag will be replaced with the original text in unstrip()
 655          *
 656          * @private
 657          */
 658         function insertStripItem( $text, &$state ) {
 659                 $rnd = $this->mUniqPrefix . '-item' . self::getRandomString();
 660                 $state->general->setPair( $rnd, $text );
 661                 return $rnd;
 662         }
 663
 664         /**
 665          * Interface with html tidy, used if $wgUseTidy = true.
 666          * If tidy isn't able to correct the markup, the original will be
 667          * returned in all its glory with a warning comment appended.
 668          *
 669          * Either the external tidy program or the in-process tidy extension
 670          * will be used depending on availability. Override the default
 671          * $wgTidyInternal setting to disable the internal if it's not working.
 672          *
 673          * @param string $text Hideous HTML input
 674          * @return string Corrected HTML output
 675          * @public
 676          * @static
 677          */
 678         function tidy( $text ) {
 679                 global $wgTidyInternal;
 680                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 681 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 682 '<head><title>test</title></head><body>'.$text.'</body></html>';
 683                 if( $wgTidyInternal ) {
 684                         $correctedtext = self::internalTidy( $wrappedtext );
 685                 } else {
 686                         $correctedtext = self::externalTidy( $wrappedtext );
 687                 }
 688                 if( is_null( $correctedtext ) ) {
 689                         wfDebug( "Tidy error detected!\n" );
 690                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 691                 }
 692                 return $correctedtext;
 693         }
 694
 695         /**
 696          * Spawn an external HTML tidy process and get corrected markup back from it.
 697          *
 698          * @private
 699          * @static
 700          */
 701         function externalTidy( $text ) {
 702                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 703                 $fname = 'Parser::externalTidy';
 704                 wfProfileIn( $fname );
 705
 706                 $cleansource = '';
 707                 $opts = ' -utf8';
 708
 709                 $descriptorspec = array(
 710                         0 => array('pipe', 'r'),
 711                         1 => array('pipe', 'w'),
 712                         2 => array('file', wfGetNull(), 'a')
 713                 );
 714                 $pipes = array();
 715                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 716                 if (is_resource($process)) {
 717                         // Theoretically, this style of communication could cause a deadlock
 718                         // here. If the stdout buffer fills up, then writes to stdin could
 719                         // block. This doesn't appear to happen with tidy, because tidy only
 720                         // writes to stdout after it's finished reading from stdin. Search
 721                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 722                         fwrite($pipes[0], $text);
 723                         fclose($pipes[0]);
 724                         while (!feof($pipes[1])) {
 725                                 $cleansource .= fgets($pipes[1], 1024);
 726                         }
 727                         fclose($pipes[1]);
 728                         proc_close($process);
 729                 }
 730
 731                 wfProfileOut( $fname );
 732
 733                 if( $cleansource == '' && $text != '') {
 734                         // Some kind of error happened, so we couldn't get the corrected text.
 735                         // Just give up; we'll use the source text and append a warning.
 736                         return null;
 737                 } else {
 738                         return $cleansource;
 739                 }
 740         }
 741
 742         /**
 743          * Use the HTML tidy PECL extension to use the tidy library in-process,
 744          * saving the overhead of spawning a new process.
 745          *
 746          * 'pear install tidy' should be able to compile the extension module.
 747          *
 748          * @private
 749          * @static
 750          */
 751         function internalTidy( $text ) {
 752                 global $wgTidyConf, $IP;
 753                 $fname = 'Parser::internalTidy';
 754                 wfProfileIn( $fname );
 755
 756                 $tidy = new tidy;
 757                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 758                 $tidy->cleanRepair();
 759                 if( $tidy->getStatus() == 2 ) {
 760                         // 2 is magic number for fatal error
 761                         // http://www.php.net/manual/en/function.tidy-get-status.php
 762                         $cleansource = null;
 763                 } else {
 764                         $cleansource = tidy_get_output( $tidy );
 765                 }
 766                 wfProfileOut( $fname );
 767                 return $cleansource;
 768         }
 769
 770         /**
 771          * parse the wiki syntax used to render tables
 772          *
 773          * @private
 774          */
 775         function doTableStuff ( $text ) {
 776                 $fname = 'Parser::doTableStuff';
 777                 wfProfileIn( $fname );
 778
 779                 $lines = explode ( "\n" , $text );
 780                 $td_history = array (); // Is currently a td tag open?
 781                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 782                 $tr_history = array (); // Is currently a tr tag open?
 783                 $tr_attributes = array (); // history of tr attributes
 784                 $has_opened_tr = array(); // Did this table open a <tr> element?
 785                 $indent_level = 0; // indent level of the table
 786                 foreach ( $lines as $key => $line )
 787                 {
 788                         $line = trim ( $line );
 789
 790                         if( $line == '' ) { // empty line, go to next line
 791                                 continue;
 792                         }
 793                         $first_character = $line{0};
 794                         $matches = array();
 795
 796                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 797                                 // First check if we are starting a new table
 798                                 $indent_level = strlen( $matches[1] );
 799
 800                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 801                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 802
 803                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 804                                 array_push ( $td_history , false );
 805                                 array_push ( $last_tag_history , '' );
 806                                 array_push ( $tr_history , false );
 807                                 array_push ( $tr_attributes , '' );
 808                                 array_push ( $has_opened_tr , false );
 809                         } else if ( count ( $td_history ) == 0 ) {
 810                                 // Don't do any of the following
 811                                 continue;
 812                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 813                                 // We are ending a table
 814                                 $line = '</table>' . substr ( $line , 2 );
 815                                 $last_tag = array_pop ( $last_tag_history );
 816
 817                                 if ( !array_pop ( $has_opened_tr ) ) {
 818                                         $line = "<tr><td></td></tr>{$line}";
 819                                 }
 820
 821                                 if ( array_pop ( $tr_history ) ) {
 822                                         $line = "</tr>{$line}";
 823                                 }
 824
 825                                 if ( array_pop ( $td_history ) ) {
 826                                         $line = "</{$last_tag}>{$line}";
 827                                 }
 828                                 array_pop ( $tr_attributes );
 829                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 830                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 831                                 // Now we have a table row
 832                                 $line = preg_replace( '#^\|-+#', '', $line );
 833
 834                                 // Whats after the tag is now only attributes
 835                                 $attributes = $this->mStripState->unstripBoth( $line );
 836                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 837                                 array_pop ( $tr_attributes );
 838                                 array_push ( $tr_attributes , $attributes );
 839
 840                                 $line = '';
 841                                 $last_tag = array_pop ( $last_tag_history );
 842                                 array_pop ( $has_opened_tr );
 843                                 array_push ( $has_opened_tr , true );
 844
 845                                 if ( array_pop ( $tr_history ) ) {
 846                                         $line = '</tr>';
 847                                 }
 848
 849                                 if ( array_pop ( $td_history ) ) {
 850                                         $line = "</{$last_tag}>{$line}";
 851                                 }
 852
 853                                 $lines[$key] = $line;
 854                                 array_push ( $tr_history , false );
 855                                 array_push ( $td_history , false );
 856                                 array_push ( $last_tag_history , '' );
 857                         }
 858                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 859                                 // This might be cell elements, td, th or captions
 860                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 861                                         $first_character = '+';
 862                                         $line = substr ( $line , 1 );
 863                                 }
 864
 865                                 $line = substr ( $line , 1 );
 866
 867                                 if ( $first_character == '!' ) {
 868                                         $line = str_replace ( '!!' , '||' , $line );
 869                                 }
 870
 871                                 // Split up multiple cells on the same line.
 872                                 // FIXME : This can result in improper nesting of tags processed
 873                                 // by earlier parser steps, but should avoid splitting up eg
 874                                 // attribute values containing literal "||".
 875                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 876
 877                                 $lines[$key] = '';
 878
 879                                 // Loop through each table cell
 880                                 foreach ( $cells as $cell )
 881                                 {
 882                                         $previous = '';
 883                                         if ( $first_character != '+' )
 884                                         {
 885                                                 $tr_after = array_pop ( $tr_attributes );
 886                                                 if ( !array_pop ( $tr_history ) ) {
 887                                                         $previous = "<tr{$tr_after}>\n";
 888                                                 }
 889                                                 array_push ( $tr_history , true );
 890                                                 array_push ( $tr_attributes , '' );
 891                                                 array_pop ( $has_opened_tr );
 892                                                 array_push ( $has_opened_tr , true );
 893                                         }
 894
 895                                         $last_tag = array_pop ( $last_tag_history );
 896
 897                                         if ( array_pop ( $td_history ) ) {
 898                                                 $previous = "</{$last_tag}>{$previous}";
 899                                         }
 900
 901                                         if ( $first_character == '|' ) {
 902                                                 $last_tag = 'td';
 903                                         } else if ( $first_character == '!' ) {
 904                                                 $last_tag = 'th';
 905                                         } else if ( $first_character == '+' ) {
 906                                                 $last_tag = 'caption';
 907                                         } else {
 908                                                 $last_tag = '';
 909                                         }
 910
 911                                         array_push ( $last_tag_history , $last_tag );
 912
 913                                         // A cell could contain both parameters and data
 914                                         $cell_data = explode ( '|' , $cell , 2 );
 915
 916                                         // Bug 553: Note that a '|' inside an invalid link should not
 917                                         // be mistaken as delimiting cell parameters
 918                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 919                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 920                                         } else if ( count ( $cell_data ) == 1 )
 921                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 922                                         else {
 923                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 924                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 925                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 926                                         }
 927
 928                                         $lines[$key] .= $cell;
 929                                         array_push ( $td_history , true );
 930                                 }
 931                         }
 932                 }
 933
 934                 // Closing open td, tr && table
 935                 while ( count ( $td_history ) > 0 )
 936                 {
 937                         if ( array_pop ( $td_history ) ) {
 938                                 $lines[] = '</td>' ;
 939                         }
 940                         if ( array_pop ( $tr_history ) ) {
 941                                 $lines[] = '</tr>' ;
 942                         }
 943                         if ( !array_pop ( $has_opened_tr ) ) {
 944                                 $lines[] = "<tr><td></td></tr>" ;
 945                         }
 946
 947                         $lines[] = '</table>' ;
 948                 }
 949
 950                 $output = implode ( "\n" , $lines ) ;
 951
 952                 // special case: don't return empty table
 953                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 954                         $output = '';
 955                 }
 956
 957                 wfProfileOut( $fname );
 958
 959                 return $output;
 960         }
 961
 962         /**
 963          * Helper function for parse() that transforms wiki markup into
 964          * HTML. Only called for $mOutputType == OT_HTML.
 965          *
 966          * @private
 967          */
 968         function internalParse( $text ) {
 969                 $args = array();
 970                 $isMain = true;
 971                 $fname = 'Parser::internalParse';
 972                 wfProfileIn( $fname );
 973
 974                 # Hook to suspend the parser in this state
 975                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 976                         wfProfileOut( $fname );
 977                         return $text ;
 978                 }
 979
 980                 # Remove <noinclude> tags and <includeonly> sections
 981                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 982                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 983                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 984
 985                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), array(), array_keys( $this->mTransparentTagHooks ) );
 986
 987                 $text = $this->replaceVariables( $text, $args );
 988                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 989
 990                 // Tables need to come after variable replacement for things to work
 991                 // properly; putting them before other transformations should keep
 992                 // exciting things like link expansions from showing up in surprising
 993                 // places.
 994                 $text = $this->doTableStuff( $text );
 995
 996                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 997
 998                 $text = $this->stripToc( $text );
 999                 $this->stripNoGallery( $text );
1000                 $text = $this->doHeadings( $text );
1001                 if($this->mOptions->getUseDynamicDates()) {
1002                         $df =& DateFormatter::getInstance();
1003                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1004                 }
1005                 $text = $this->doAllQuotes( $text );
1006                 $text = $this->replaceInternalLinks( $text );
1007                 $text = $this->replaceExternalLinks( $text );
1008
1009                 # replaceInternalLinks may sometimes leave behind
1010                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1011                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1012
1013                 $text = $this->doMagicLinks( $text );
1014                 $text = $this->formatHeadings( $text, $isMain );
1015
1016                 wfProfileOut( $fname );
1017                 return $text;
1018         }
1019
1020         /**
1021          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1022          * magic external links.
1023          *
1024          * @private
1025          */
1026         function &doMagicLinks( &$text ) {
1027                 wfProfileIn( __METHOD__ );
1028                 $text = preg_replace_callback(
1029                         '!(?:                           # Start cases
1030                             <a.*?</a> |                 # Skip link text
1031                             <.*?> |                     # Skip stuff inside HTML elements
1032                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1033                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1034                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1035                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1036                                      [0-9Xx]                 # check digit
1037                                    \b)
1038                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1039                 wfProfileOut( __METHOD__ );
1040                 return $text;
1041         }
1042
1043         function magicLinkCallback( $m ) {
1044                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1045                         # Skip HTML element
1046                         return $m[0];
1047                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1048                         $isbn = $m[2];
1049                         $num = strtr( $isbn, array(
1050                                 '-' => '',
1051                                 ' ' => '',
1052                                 'x' => 'X',
1053                         ));
1054                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1055                         $text = '<a href="' .
1056                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1057                                 "\" class=\"internal\">ISBN $isbn</a>";
1058                 } else {
1059                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1060                                 $keyword = 'RFC';
1061                                 $urlmsg = 'rfcurl';
1062                                 $id = $m[1];
1063                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1064                                 $keyword = 'PMID';
1065                                 $urlmsg = 'pubmedurl';
1066                                 $id = $m[1];
1067                         } else {
1068                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1069                                         substr($m[0], 0, 20 ) . '"' );
1070                         }
1071
1072                         $url = wfMsg( $urlmsg, $id);
1073                         $sk = $this->mOptions->getSkin();
1074                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1075                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1076                 }
1077                 return $text;
1078         }
1079
1080         /**
1081          * Parse headers and return html
1082          *
1083          * @private
1084          */
1085         function doHeadings( $text ) {
1086                 $fname = 'Parser::doHeadings';
1087                 wfProfileIn( $fname );
1088                 for ( $i = 6; $i >= 1; --$i ) {
1089                         $h = str_repeat( '=', $i );
1090                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1091                           "<h{$i}>\\1</h{$i}>\\2", $text );
1092                 }
1093                 wfProfileOut( $fname );
1094                 return $text;
1095         }
1096
1097         /**
1098          * Replace single quotes with HTML markup
1099          * @private
1100          * @return string the altered text
1101          */
1102         function doAllQuotes( $text ) {
1103                 $fname = 'Parser::doAllQuotes';
1104                 wfProfileIn( $fname );
1105                 $outtext = '';
1106                 $lines = explode( "\n", $text );
1107                 foreach ( $lines as $line ) {
1108                         $outtext .= $this->doQuotes ( $line ) . "\n";
1109                 }
1110                 $outtext = substr($outtext, 0,-1);
1111                 wfProfileOut( $fname );
1112                 return $outtext;
1113         }
1114
1115         /**
1116          * Helper function for doAllQuotes()
1117          */
1118         public function doQuotes( $text ) {
1119                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1120                 if ( count( $arr ) == 1 )
1121                         return $text;
1122                 else
1123                 {
1124                         # First, do some preliminary work. This may shift some apostrophes from
1125                         # being mark-up to being text. It also counts the number of occurrences
1126                         # of bold and italics mark-ups.
1127                         $i = 0;
1128                         $numbold = 0;
1129                         $numitalics = 0;
1130                         foreach ( $arr as $r )
1131                         {
1132                                 if ( ( $i % 2 ) == 1 )
1133                                 {
1134                                         # If there are ever four apostrophes, assume the first is supposed to
1135                                         # be text, and the remaining three constitute mark-up for bold text.
1136                                         if ( strlen( $arr[$i] ) == 4 )
1137                                         {
1138                                                 $arr[$i-1] .= "'";
1139                                                 $arr[$i] = "'''";
1140                                         }
1141                                         # If there are more than 5 apostrophes in a row, assume they're all
1142                                         # text except for the last 5.
1143                                         else if ( strlen( $arr[$i] ) > 5 )
1144                                         {
1145                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1146                                                 $arr[$i] = "'''''";
1147                                         }
1148                                         # Count the number of occurrences of bold and italics mark-ups.
1149                                         # We are not counting sequences of five apostrophes.
1150                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1151                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1152                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1153                                 }
1154                                 $i++;
1155                         }
1156
1157                         # If there is an odd number of both bold and italics, it is likely
1158                         # that one of the bold ones was meant to be an apostrophe followed
1159                         # by italics. Which one we cannot know for certain, but it is more
1160                         # likely to be one that has a single-letter word before it.
1161                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1162                         {
1163                                 $i = 0;
1164                                 $firstsingleletterword = -1;
1165                                 $firstmultiletterword = -1;
1166                                 $firstspace = -1;
1167                                 foreach ( $arr as $r )
1168                                 {
1169                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1170                                         {
1171                                                 $x1 = substr ($arr[$i-1], -1);
1172                                                 $x2 = substr ($arr[$i-1], -2, 1);
1173                                                 if ($x1 == ' ') {
1174                                                         if ($firstspace == -1) $firstspace = $i;
1175                                                 } else if ($x2 == ' ') {
1176                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1177                                                 } else {
1178                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1179                                                 }
1180                                         }
1181                                         $i++;
1182                                 }
1183
1184                                 # If there is a single-letter word, use it!
1185                                 if ($firstsingleletterword > -1)
1186                                 {
1187                                         $arr [ $firstsingleletterword ] = "''";
1188                                         $arr [ $firstsingleletterword-1 ] .= "'";
1189                                 }
1190                                 # If not, but there's a multi-letter word, use that one.
1191                                 else if ($firstmultiletterword > -1)
1192                                 {
1193                                         $arr [ $firstmultiletterword ] = "''";
1194                                         $arr [ $firstmultiletterword-1 ] .= "'";
1195                                 }
1196                                 # ... otherwise use the first one that has neither.
1197                                 # (notice that it is possible for all three to be -1 if, for example,
1198                                 # there is only one pentuple-apostrophe in the line)
1199                                 else if ($firstspace > -1)
1200                                 {
1201                                         $arr [ $firstspace ] = "''";
1202                                         $arr [ $firstspace-1 ] .= "'";
1203                                 }
1204                         }
1205
1206                         # Now let's actually convert our apostrophic mush to HTML!
1207                         $output = '';
1208                         $buffer = '';
1209                         $state = '';
1210                         $i = 0;
1211                         foreach ($arr as $r)
1212                         {
1213                                 if (($i % 2) == 0)
1214                                 {
1215                                         if ($state == 'both')
1216                                                 $buffer .= $r;
1217                                         else
1218                                                 $output .= $r;
1219                                 }
1220                                 else
1221                                 {
1222                                         if (strlen ($r) == 2)
1223                                         {
1224                                                 if ($state == 'i')
1225                                                 { $output .= '</i>'; $state = ''; }
1226                                                 else if ($state == 'bi')
1227                                                 { $output .= '</i>'; $state = 'b'; }
1228                                                 else if ($state == 'ib')
1229                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1230                                                 else if ($state == 'both')
1231                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1232                                                 else # $state can be 'b' or ''
1233                                                 { $output .= '<i>'; $state .= 'i'; }
1234                                         }
1235                                         else if (strlen ($r) == 3)
1236                                         {
1237                                                 if ($state == 'b')
1238                                                 { $output .= '</b>'; $state = ''; }
1239                                                 else if ($state == 'bi')
1240                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1241                                                 else if ($state == 'ib')
1242                                                 { $output .= '</b>'; $state = 'i'; }
1243                                                 else if ($state == 'both')
1244                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1245                                                 else # $state can be 'i' or ''
1246                                                 { $output .= '<b>'; $state .= 'b'; }
1247                                         }
1248                                         else if (strlen ($r) == 5)
1249                                         {
1250                                                 if ($state == 'b')
1251                                                 { $output .= '</b><i>'; $state = 'i'; }
1252                                                 else if ($state == 'i')
1253                                                 { $output .= '</i><b>'; $state = 'b'; }
1254                                                 else if ($state == 'bi')
1255                                                 { $output .= '</i></b>'; $state = ''; }
1256                                                 else if ($state == 'ib')
1257                                                 { $output .= '</b></i>'; $state = ''; }
1258                                                 else if ($state == 'both')
1259                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1260                                                 else # ($state == '')
1261                                                 { $buffer = ''; $state = 'both'; }
1262                                         }
1263                                 }
1264                                 $i++;
1265                         }
1266                         # Now close all remaining tags.  Notice that the order is important.
1267                         if ($state == 'b' || $state == 'ib')
1268                                 $output .= '</b>';
1269                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1270                                 $output .= '</i>';
1271                         if ($state == 'bi')
1272                                 $output .= '</b>';
1273                         # There might be lonely ''''', so make sure we have a buffer
1274                         if ($state == 'both' && $buffer)
1275                                 $output .= '<b><i>'.$buffer.'</i></b>';
1276                         return $output;
1277                 }
1278         }
1279
1280         /**
1281          * Replace external links
1282          *
1283          * Note: this is all very hackish and the order of execution matters a lot.
1284          * Make sure to run maintenance/parserTests.php if you change this code.
1285          *
1286          * @private
1287          */
1288         function replaceExternalLinks( $text ) {
1289                 global $wgContLang;
1290                 $fname = 'Parser::replaceExternalLinks';
1291                 wfProfileIn( $fname );
1292
1293                 $sk = $this->mOptions->getSkin();
1294
1295                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1296
1297                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1298
1299                 $i = 0;
1300                 while ( $i<count( $bits ) ) {
1301                         $url = $bits[$i++];
1302                         $protocol = $bits[$i++];
1303                         $text = $bits[$i++];
1304                         $trail = $bits[$i++];
1305
1306                         # The characters '<' and '>' (which were escaped by
1307                         # removeHTMLtags()) should not be included in
1308                         # URLs, per RFC 2396.
1309                         $m2 = array();
1310                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1311                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1312                                 $url = substr($url, 0, $m2[0][1]);
1313                         }
1314
1315                         # If the link text is an image URL, replace it with an <img> tag
1316                         # This happened by accident in the original parser, but some people used it extensively
1317                         $img = $this->maybeMakeExternalImage( $text );
1318                         if ( $img !== false ) {
1319                                 $text = $img;
1320                         }
1321
1322                         $dtrail = '';
1323
1324                         # Set linktype for CSS - if URL==text, link is essentially free
1325                         $linktype = ($text == $url) ? 'free' : 'text';
1326
1327                         # No link text, e.g. [http://domain.tld/some.link]
1328                         if ( $text == '' ) {
1329                                 # Autonumber if allowed. See bug #5918
1330                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1331                                         $text = '[' . ++$this->mAutonumber . ']';
1332                                         $linktype = 'autonumber';
1333                                 } else {
1334                                         # Otherwise just use the URL
1335                                         $text = htmlspecialchars( $url );
1336                                         $linktype = 'free';
1337                                 }
1338                         } else {
1339                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1340                                 # Check for trail
1341                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1342                         }
1343
1344                         $text = $wgContLang->markNoConversion($text);
1345
1346                         $url = Sanitizer::cleanUrl( $url );
1347
1348                         # Process the trail (i.e. everything after this link up until start of the next link),
1349                         # replacing any non-bracketed links
1350                         $trail = $this->replaceFreeExternalLinks( $trail );
1351
1352                         # Use the encoded URL
1353                         # This means that users can paste URLs directly into the text
1354                         # Funny characters like &ouml; aren't valid in URLs anyway
1355                         # This was changed in August 2004
1356                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1357
1358                         # Register link in the output object.
1359                         # Replace unnecessary URL escape codes with the referenced character
1360                         # This prevents spammers from hiding links from the filters
1361                         $pasteurized = self::replaceUnusualEscapes( $url );
1362                         $this->mOutput->addExternalLink( $pasteurized );
1363                 }
1364
1365                 wfProfileOut( $fname );
1366                 return $s;
1367         }
1368
1369         /**
1370          * Replace anything that looks like a URL with a link
1371          * @private
1372          */
1373         function replaceFreeExternalLinks( $text ) {
1374                 global $wgContLang;
1375                 $fname = 'Parser::replaceFreeExternalLinks';
1376                 wfProfileIn( $fname );
1377
1378                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1379                 $s = array_shift( $bits );
1380                 $i = 0;
1381
1382                 $sk = $this->mOptions->getSkin();
1383
1384                 while ( $i < count( $bits ) ){
1385                         $protocol = $bits[$i++];
1386                         $remainder = $bits[$i++];
1387
1388                         $m = array();
1389                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1390                                 # Found some characters after the protocol that look promising
1391                                 $url = $protocol . $m[1];
1392                                 $trail = $m[2];
1393
1394                                 # special case: handle urls as url args:
1395                                 # http://www.example.com/foo?=http://www.example.com/bar
1396                                 if(strlen($trail) == 0 &&
1397                                         isset($bits[$i]) &&
1398                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1399                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1400                                 {
1401                                         # add protocol, arg
1402                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1403                                         $i += 2;
1404                                         $trail = $m[2];
1405                                 }
1406
1407                                 # The characters '<' and '>' (which were escaped by
1408                                 # removeHTMLtags()) should not be included in
1409                                 # URLs, per RFC 2396.
1410                                 $m2 = array();
1411                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1412                                         $trail = substr($url, $m2[0][1]) . $trail;
1413                                         $url = substr($url, 0, $m2[0][1]);
1414                                 }
1415
1416                                 # Move trailing punctuation to $trail
1417                                 $sep = ',;\.:!?';
1418                                 # If there is no left bracket, then consider right brackets fair game too
1419                                 if ( strpos( $url, '(' ) === false ) {
1420                                         $sep .= ')';
1421                                 }
1422
1423                                 $numSepChars = strspn( strrev( $url ), $sep );
1424                                 if ( $numSepChars ) {
1425                                         $trail = substr( $url, -$numSepChars ) . $trail;
1426                                         $url = substr( $url, 0, -$numSepChars );
1427                                 }
1428
1429                                 $url = Sanitizer::cleanUrl( $url );
1430
1431                                 # Is this an external image?
1432                                 $text = $this->maybeMakeExternalImage( $url );
1433                                 if ( $text === false ) {
1434                                         # Not an image, make a link
1435                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1436                                         # Register it in the output object...
1437                                         # Replace unnecessary URL escape codes with their equivalent characters
1438                                         $pasteurized = self::replaceUnusualEscapes( $url );
1439                                         $this->mOutput->addExternalLink( $pasteurized );
1440                                 }
1441                                 $s .= $text . $trail;
1442                         } else {
1443                                 $s .= $protocol . $remainder;
1444                         }
1445                 }
1446                 wfProfileOut( $fname );
1447                 return $s;
1448         }
1449
1450         /**
1451          * Replace unusual URL escape codes with their equivalent characters
1452          * @param string
1453          * @return string
1454          * @static
1455          * @todo  This can merge genuinely required bits in the path or query string,
1456          *        breaking legit URLs. A proper fix would treat the various parts of
1457          *        the URL differently; as a workaround, just use the output for
1458          *        statistical records, not for actual linking/output.
1459          */
1460         static function replaceUnusualEscapes( $url ) {
1461                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1462                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1463         }
1464
1465         /**
1466          * Callback function used in replaceUnusualEscapes().
1467          * Replaces unusual URL escape codes with their equivalent character
1468          * @static
1469          * @private
1470          */
1471         private static function replaceUnusualEscapesCallback( $matches ) {
1472                 $char = urldecode( $matches[0] );
1473                 $ord = ord( $char );
1474                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1475                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1476                         // No, shouldn't be escaped
1477                         return $char;
1478                 } else {
1479                         // Yes, leave it escaped
1480                         return $matches[0];
1481                 }
1482         }
1483
1484         /**
1485          * make an image if it's allowed, either through the global
1486          * option or through the exception
1487          * @private
1488          */
1489         function maybeMakeExternalImage( $url ) {
1490                 $sk = $this->mOptions->getSkin();
1491                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1492                 $imagesexception = !empty($imagesfrom);
1493                 $text = false;
1494                 if ( $this->mOptions->getAllowExternalImages()
1495                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1496                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1497                                 # Image found
1498                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1499                         }
1500                 }
1501                 return $text;
1502         }
1503
1504         /**
1505          * Process [[ ]] wikilinks
1506          *
1507          * @private
1508          */
1509         function replaceInternalLinks( $s ) {
1510                 global $wgContLang;
1511                 static $fname = 'Parser::replaceInternalLinks' ;
1512
1513                 wfProfileIn( $fname );
1514
1515                 wfProfileIn( $fname.'-setup' );
1516                 static $tc = FALSE;
1517                 # the % is needed to support urlencoded titles as well
1518                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1519
1520                 $sk = $this->mOptions->getSkin();
1521
1522                 #split the entire text string on occurences of [[
1523                 $a = explode( '[[', ' ' . $s );
1524                 #get the first element (all text up to first [[), and remove the space we added
1525                 $s = array_shift( $a );
1526                 $s = substr( $s, 1 );
1527
1528                 # Match a link having the form [[namespace:link|alternate]]trail
1529                 static $e1 = FALSE;
1530                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1531                 # Match cases where there is no "]]", which might still be images
1532                 static $e1_img = FALSE;
1533                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1534                 # Match the end of a line for a word that's not followed by whitespace,
1535                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1536                 $e2 = wfMsgForContent( 'linkprefix' );
1537
1538                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1539                 if( is_null( $this->mTitle ) ) {
1540                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1541                 }
1542                 $nottalk = !$this->mTitle->isTalkPage();
1543
1544                 if ( $useLinkPrefixExtension ) {
1545                         $m = array();
1546                         if ( preg_match( $e2, $s, $m ) ) {
1547                                 $first_prefix = $m[2];
1548                         } else {
1549                                 $first_prefix = false;
1550                         }
1551                 } else {
1552                         $prefix = '';
1553                 }
1554
1555                 if($wgContLang->hasVariants()) {
1556                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1557                 } else {
1558                         $selflink = array($this->mTitle->getPrefixedText());
1559                 }
1560                 $useSubpages = $this->areSubpagesAllowed();
1561                 wfProfileOut( $fname.'-setup' );
1562
1563                 # Loop for each link
1564                 for ($k = 0; isset( $a[$k] ); $k++) {
1565                         $line = $a[$k];
1566                         if ( $useLinkPrefixExtension ) {
1567                                 wfProfileIn( $fname.'-prefixhandling' );
1568                                 if ( preg_match( $e2, $s, $m ) ) {
1569                                         $prefix = $m[2];
1570                                         $s = $m[1];
1571                                 } else {
1572                                         $prefix='';
1573                                 }
1574                                 # first link
1575                                 if($first_prefix) {
1576                                         $prefix = $first_prefix;
1577                                         $first_prefix = false;
1578                                 }
1579                                 wfProfileOut( $fname.'-prefixhandling' );
1580                         }
1581
1582                         $might_be_img = false;
1583
1584                         wfProfileIn( "$fname-e1" );
1585                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1586                                 $text = $m[2];
1587                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1588                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1589                                 # the real problem is with the $e1 regex
1590                                 # See bug 1300.
1591                                 #
1592                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1593                                 # and no image is in sight. See bug 2095.
1594                                 #
1595                                 if( $text !== '' &&
1596                                         substr( $m[3], 0, 1 ) === ']' &&
1597                                         strpos($text, '[') !== false
1598                                 )
1599                                 {
1600                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1601                                         $m[3] = substr( $m[3], 1 );
1602                                 }
1603                                 # fix up urlencoded title texts
1604                                 if( strpos( $m[1], '%' ) !== false ) {
1605                                         # Should anchors '#' also be rejected?
1606                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1607                                 }
1608                                 $trail = $m[3];
1609                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1610                                 $might_be_img = true;
1611                                 $text = $m[2];
1612                                 if ( strpos( $m[1], '%' ) !== false ) {
1613                                         $m[1] = urldecode($m[1]);
1614                                 }
1615                                 $trail = "";
1616                         } else { # Invalid form; output directly
1617                                 $s .= $prefix . '[[' . $line ;
1618                                 wfProfileOut( "$fname-e1" );
1619                                 continue;
1620                         }
1621                         wfProfileOut( "$fname-e1" );
1622                         wfProfileIn( "$fname-misc" );
1623
1624                         # Don't allow internal links to pages containing
1625                         # PROTO: where PROTO is a valid URL protocol; these
1626                         # should be external links.
1627                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1628                                 $s .= $prefix . '[[' . $line ;
1629                                 continue;
1630                         }
1631
1632                         # Make subpage if necessary
1633                         if( $useSubpages ) {
1634                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1635                         } else {
1636                                 $link = $m[1];
1637                         }
1638
1639                         $noforce = (substr($m[1], 0, 1) != ':');
1640                         if (!$noforce) {
1641                                 # Strip off leading ':'
1642                                 $link = substr($link, 1);
1643                         }
1644
1645                         wfProfileOut( "$fname-misc" );
1646                         wfProfileIn( "$fname-title" );
1647                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1648                         if( !$nt ) {
1649                                 $s .= $prefix . '[[' . $line;
1650                                 wfProfileOut( "$fname-title" );
1651                                 continue;
1652                         }
1653
1654                         $ns = $nt->getNamespace();
1655                         $iw = $nt->getInterWiki();
1656                         wfProfileOut( "$fname-title" );
1657
1658                         if ($might_be_img) { # if this is actually an invalid link
1659                                 wfProfileIn( "$fname-might_be_img" );
1660                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1661                                         $found = false;
1662                                         while (isset ($a[$k+1]) ) {
1663                                                 #look at the next 'line' to see if we can close it there
1664                                                 $spliced = array_splice( $a, $k + 1, 1 );
1665                                                 $next_line = array_shift( $spliced );
1666                                                 $m = explode( ']]', $next_line, 3 );
1667                                                 if ( count( $m ) == 3 ) {
1668                                                         # the first ]] closes the inner link, the second the image
1669                                                         $found = true;
1670                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1671                                                         $trail = $m[2];
1672                                                         break;
1673                                                 } elseif ( count( $m ) == 2 ) {
1674                                                         #if there's exactly one ]] that's fine, we'll keep looking
1675                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1676                                                 } else {
1677                                                         #if $next_line is invalid too, we need look no further
1678                                                         $text .= '[[' . $next_line;
1679                                                         break;
1680                                                 }
1681                                         }
1682                                         if ( !$found ) {
1683                                                 # we couldn't find the end of this imageLink, so output it raw
1684                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1685                                                 $text = $this->replaceInternalLinks($text);
1686                                                 $s .= "{$prefix}[[$link|$text";
1687                                                 # note: no $trail, because without an end, there *is* no trail
1688                                                 wfProfileOut( "$fname-might_be_img" );
1689                                                 continue;
1690                                         }
1691                                 } else { #it's not an image, so output it raw
1692                                         $s .= "{$prefix}[[$link|$text";
1693                                         # note: no $trail, because without an end, there *is* no trail
1694                                         wfProfileOut( "$fname-might_be_img" );
1695                                         continue;
1696                                 }
1697                                 wfProfileOut( "$fname-might_be_img" );
1698                         }
1699
1700                         $wasblank = ( '' == $text );
1701                         if( $wasblank ) $text = $link;
1702
1703                         # Link not escaped by : , create the various objects
1704                         if( $noforce ) {
1705
1706                                 # Interwikis
1707                                 wfProfileIn( "$fname-interwiki" );
1708                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1709                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1710                                         $s = rtrim($s . $prefix);
1711                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1712                                         wfProfileOut( "$fname-interwiki" );
1713                                         continue;
1714                                 }
1715                                 wfProfileOut( "$fname-interwiki" );
1716
1717                                 if ( $ns == NS_IMAGE ) {
1718                                         wfProfileIn( "$fname-image" );
1719                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1720                                                 # recursively parse links inside the image caption
1721                                                 # actually, this will parse them in any other parameters, too,
1722                                                 # but it might be hard to fix that, and it doesn't matter ATM
1723                                                 $text = $this->replaceExternalLinks($text);
1724                                                 $text = $this->replaceInternalLinks($text);
1725
1726                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1727                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1728                                                 $this->mOutput->addImage( $nt->getDBkey() );
1729
1730                                                 wfProfileOut( "$fname-image" );
1731                                                 continue;
1732                                         } else {
1733                                                 # We still need to record the image's presence on the page
1734                                                 $this->mOutput->addImage( $nt->getDBkey() );
1735                                         }
1736                                         wfProfileOut( "$fname-image" );
1737
1738                                 }
1739
1740                                 if ( $ns == NS_CATEGORY ) {
1741                                         wfProfileIn( "$fname-category" );
1742                                         $s = rtrim($s . "\n"); # bug 87
1743
1744                                         if ( $wasblank ) {
1745                                                 $sortkey = $this->getDefaultSort();
1746                                         } else {
1747                                                 $sortkey = $text;
1748                                         }
1749                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1750                                         $sortkey = str_replace( "\n", '', $sortkey );
1751                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1752                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1753
1754                                         /**
1755                                          * Strip the whitespace Category links produce, see bug 87
1756                                          * @todo We might want to use trim($tmp, "\n") here.
1757                                          */
1758                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1759
1760                                         wfProfileOut( "$fname-category" );
1761                                         continue;
1762                                 }
1763                         }
1764
1765                         # Self-link checking
1766                         if( $nt->getFragment() === '' ) {
1767                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1768                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1769                                         continue;
1770                                 }
1771                         }
1772
1773                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1774                         if( $ns == NS_MEDIA ) {
1775                                 # Give extensions a chance to select the file revision for us
1776                                 $skip = $time = false;
1777                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1778                                 if ( $skip ) {
1779                                         $link = $sk->makeLinkObj( $nt );
1780                                 } else {
1781                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1782                                 }
1783                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1784                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1785                                 $this->mOutput->addImage( $nt->getDBkey() );
1786                                 continue;
1787                         } elseif( $ns == NS_SPECIAL ) {
1788                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1789                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1790                                 } else {
1791                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1792                                 }
1793                                 continue;
1794                         } elseif( $ns == NS_IMAGE ) {
1795                                 $img = wfFindFile( $nt );
1796                                 if( $img ) {
1797                                         // Force a blue link if the file exists; may be a remote
1798                                         // upload on the shared repository, and we want to see its
1799                                         // auto-generated page.
1800                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1801                                         $this->mOutput->addLink( $nt );
1802                                         continue;
1803                                 }
1804                         }
1805                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1806                 }
1807                 wfProfileOut( $fname );
1808                 return $s;
1809         }
1810
1811         /**
1812          * Make a link placeholder. The text returned can be later resolved to a real link with
1813          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1814          * parsing of interwiki links, and secondly to allow all existence checks and
1815          * article length checks (for stub links) to be bundled into a single query.
1816          *
1817          */
1818         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1819                 wfProfileIn( __METHOD__ );
1820                 if ( ! is_object($nt) ) {
1821                         # Fail gracefully
1822                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1823                 } else {
1824                         # Separate the link trail from the rest of the link
1825                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1826
1827                         if ( $nt->isExternal() ) {
1828                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1829                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1830                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1831                         } else {
1832                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1833                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1834                                 $this->mLinkHolders['queries'][] = $query;
1835                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1836                                 $this->mLinkHolders['titles'][] = $nt;
1837
1838                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1839                         }
1840                 }
1841                 wfProfileOut( __METHOD__ );
1842                 return $retVal;
1843         }
1844
1845         /**
1846          * Render a forced-blue link inline; protect against double expansion of
1847          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1848          * Since this little disaster has to split off the trail text to avoid
1849          * breaking URLs in the following text without breaking trails on the
1850          * wiki links, it's been made into a horrible function.
1851          *
1852          * @param Title $nt
1853          * @param string $text
1854          * @param string $query
1855          * @param string $trail
1856          * @param string $prefix
1857          * @return string HTML-wikitext mix oh yuck
1858          */
1859         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1860                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1861                 $sk = $this->mOptions->getSkin();
1862                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1863                 return $this->armorLinks( $link ) . $trail;
1864         }
1865
1866         /**
1867          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1868          * going to go through further parsing steps before inline URL expansion.
1869          *
1870          * In particular this is important when using action=render, which causes
1871          * full URLs to be included.
1872          *
1873          * Oh man I hate our multi-layer parser!
1874          *
1875          * @param string more-or-less HTML
1876          * @return string less-or-more HTML with NOPARSE bits
1877          */
1878         function armorLinks( $text ) {
1879                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1880                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1881         }
1882
1883         /**
1884          * Return true if subpage links should be expanded on this page.
1885          * @return bool
1886          */
1887         function areSubpagesAllowed() {
1888                 # Some namespaces don't allow subpages
1889                 global $wgNamespacesWithSubpages;
1890                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1891         }
1892
1893         /**
1894          * Handle link to subpage if necessary
1895          * @param string $target the source of the link
1896          * @param string &$text the link text, modified as necessary
1897          * @return string the full name of the link
1898          * @private
1899          */
1900         function maybeDoSubpageLink($target, &$text) {
1901                 # Valid link forms:
1902                 # Foobar -- normal
1903                 # :Foobar -- override special treatment of prefix (images, language links)
1904                 # /Foobar -- convert to CurrentPage/Foobar
1905                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1906                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1907                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1908
1909                 $fname = 'Parser::maybeDoSubpageLink';
1910                 wfProfileIn( $fname );
1911                 $ret = $target; # default return value is no change
1912
1913                 # Some namespaces don't allow subpages,
1914                 # so only perform processing if subpages are allowed
1915                 if( $this->areSubpagesAllowed() ) {
1916                         $hash = strpos( $target, '#' );
1917                         if( $hash !== false ) {
1918                                 $suffix = substr( $target, $hash );
1919                                 $target = substr( $target, 0, $hash );
1920                         } else {
1921                                 $suffix = '';
1922                         }
1923                         # bug 7425
1924                         $target = trim( $target );
1925                         # Look at the first character
1926                         if( $target != '' && $target{0} == '/' ) {
1927                                 # / at end means we don't want the slash to be shown
1928                                 $m = array();
1929                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1930                                 if( $trailingSlashes ) {
1931                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1932                                 } else {
1933                                         $noslash = substr( $target, 1 );
1934                                 }
1935
1936                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1937                                 if( '' === $text ) {
1938                                         $text = $target . $suffix;
1939                                 } # this might be changed for ugliness reasons
1940                         } else {
1941                                 # check for .. subpage backlinks
1942                                 $dotdotcount = 0;
1943                                 $nodotdot = $target;
1944                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1945                                         ++$dotdotcount;
1946                                         $nodotdot = substr( $nodotdot, 3 );
1947                                 }
1948                                 if($dotdotcount > 0) {
1949                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1950                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1951                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1952                                                 # / at the end means don't show full path
1953                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1954                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1955                                                         if( '' === $text ) {
1956                                                                 $text = $nodotdot . $suffix;
1957                                                         }
1958                                                 }
1959                                                 $nodotdot = trim( $nodotdot );
1960                                                 if( $nodotdot != '' ) {
1961                                                         $ret .= '/' . $nodotdot;
1962                                                 }
1963                                                 $ret .= $suffix;
1964                                         }
1965                                 }
1966                         }
1967                 }
1968
1969                 wfProfileOut( $fname );
1970                 return $ret;
1971         }
1972
1973         /**#@+
1974          * Used by doBlockLevels()
1975          * @private
1976          */
1977         /* private */ function closeParagraph() {
1978                 $result = '';
1979                 if ( '' != $this->mLastSection ) {
1980                         $result = '</' . $this->mLastSection  . ">\n";
1981                 }
1982                 $this->mInPre = false;
1983                 $this->mLastSection = '';
1984                 return $result;
1985         }
1986         # getCommon() returns the length of the longest common substring
1987         # of both arguments, starting at the beginning of both.
1988         #
1989         /* private */ function getCommon( $st1, $st2 ) {
1990                 $fl = strlen( $st1 );
1991                 $shorter = strlen( $st2 );
1992                 if ( $fl < $shorter ) { $shorter = $fl; }
1993
1994                 for ( $i = 0; $i < $shorter; ++$i ) {
1995                         if ( $st1{$i} != $st2{$i} ) { break; }
1996                 }
1997                 return $i;
1998         }
1999         # These next three functions open, continue, and close the list
2000         # element appropriate to the prefix character passed into them.
2001         #
2002         /* private */ function openList( $char ) {
2003                 $result = $this->closeParagraph();
2004
2005                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2006                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2007                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2008                 else if ( ';' == $char ) {
2009                         $result .= '<dl><dt>';
2010                         $this->mDTopen = true;
2011                 }
2012                 else { $result = '<!-- ERR 1 -->'; }
2013
2014                 return $result;
2015         }
2016
2017         /* private */ function nextItem( $char ) {
2018                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2019                 else if ( ':' == $char || ';' == $char ) {
2020                         $close = '</dd>';
2021                         if ( $this->mDTopen ) { $close = '</dt>'; }
2022                         if ( ';' == $char ) {
2023                                 $this->mDTopen = true;
2024                                 return $close . '<dt>';
2025                         } else {
2026                                 $this->mDTopen = false;
2027                                 return $close . '<dd>';
2028                         }
2029                 }
2030                 return '<!-- ERR 2 -->';
2031         }
2032
2033         /* private */ function closeList( $char ) {
2034                 if ( '*' == $char ) { $text = '</li></ul>'; }
2035                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2036                 else if ( ':' == $char ) {
2037                         if ( $this->mDTopen ) {
2038                                 $this->mDTopen = false;
2039                                 $text = '</dt></dl>';
2040                         } else {
2041                                 $text = '</dd></dl>';
2042                         }
2043                 }
2044                 else {  return '<!-- ERR 3 -->'; }
2045                 return $text."\n";
2046         }
2047         /**#@-*/
2048
2049         /**
2050          * Make lists from lines starting with ':', '*', '#', etc.
2051          *
2052          * @private
2053          * @return string the lists rendered as HTML
2054          */
2055         function doBlockLevels( $text, $linestart ) {
2056                 $fname = 'Parser::doBlockLevels';
2057                 wfProfileIn( $fname );
2058
2059                 # Parsing through the text line by line.  The main thing
2060                 # happening here is handling of block-level elements p, pre,
2061                 # and making lists from lines starting with * # : etc.
2062                 #
2063                 $textLines = explode( "\n", $text );
2064
2065                 $lastPrefix = $output = '';
2066                 $this->mDTopen = $inBlockElem = false;
2067                 $prefixLength = 0;
2068                 $paragraphStack = false;
2069
2070                 if ( !$linestart ) {
2071                         $output .= array_shift( $textLines );
2072                 }
2073                 foreach ( $textLines as $oLine ) {
2074                         $lastPrefixLength = strlen( $lastPrefix );
2075                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2076                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2077                         if ( !$this->mInPre ) {
2078                                 # Multiple prefixes may abut each other for nested lists.
2079                                 $prefixLength = strspn( $oLine, '*#:;' );
2080                                 $pref = substr( $oLine, 0, $prefixLength );
2081
2082                                 # eh?
2083                                 $pref2 = str_replace( ';', ':', $pref );
2084                                 $t = substr( $oLine, $prefixLength );
2085                                 $this->mInPre = !empty($preOpenMatch);
2086                         } else {
2087                                 # Don't interpret any other prefixes in preformatted text
2088                                 $prefixLength = 0;
2089                                 $pref = $pref2 = '';
2090                                 $t = $oLine;
2091                         }
2092
2093                         # List generation
2094                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2095                                 # Same as the last item, so no need to deal with nesting or opening stuff
2096                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2097                                 $paragraphStack = false;
2098
2099                                 if ( substr( $pref, -1 ) == ';') {
2100                                         # The one nasty exception: definition lists work like this:
2101                                         # ; title : definition text
2102                                         # So we check for : in the remainder text to split up the
2103                                         # title and definition, without b0rking links.
2104                                         $term = $t2 = '';
2105                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2106                                                 $t = $t2;
2107                                                 $output .= $term . $this->nextItem( ':' );
2108                                         }
2109                                 }
2110                         } elseif( $prefixLength || $lastPrefixLength ) {
2111                                 # Either open or close a level...
2112                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2113                                 $paragraphStack = false;
2114
2115                                 while( $commonPrefixLength < $lastPrefixLength ) {
2116                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2117                                         --$lastPrefixLength;
2118                                 }
2119                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2120                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2121                                 }
2122                                 while ( $prefixLength > $commonPrefixLength ) {
2123                                         $char = substr( $pref, $commonPrefixLength, 1 );
2124                                         $output .= $this->openList( $char );
2125
2126                                         if ( ';' == $char ) {
2127                                                 # FIXME: This is dupe of code above
2128                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2129                                                         $t = $t2;
2130                                                         $output .= $term . $this->nextItem( ':' );
2131                                                 }
2132                                         }
2133                                         ++$commonPrefixLength;
2134                                 }
2135                                 $lastPrefix = $pref2;
2136                         }
2137                         if( 0 == $prefixLength ) {
2138                                 wfProfileIn( "$fname-paragraph" );
2139                                 # No prefix (not in list)--go to paragraph mode
2140                                 // XXX: use a stack for nestable elements like span, table and div
2141                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2142                                 $closematch = preg_match(
2143                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2144                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2145                                 if ( $openmatch or $closematch ) {
2146                                         $paragraphStack = false;
2147                                         # TODO bug 5718: paragraph closed
2148                                         $output .= $this->closeParagraph();
2149                                         if ( $preOpenMatch and !$preCloseMatch ) {
2150                                                 $this->mInPre = true;
2151                                         }
2152                                         if ( $closematch ) {
2153                                                 $inBlockElem = false;
2154                                         } else {
2155                                                 $inBlockElem = true;
2156                                         }
2157                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2158                                         if ( '' != $t and ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2159                                                 // pre
2160                                                 if ($this->mLastSection != 'pre') {
2161                                                         $paragraphStack = false;
2162                                                         $output .= $this->closeParagraph().'<pre>';
2163                                                         $this->mLastSection = 'pre';
2164                                                 }
2165                                                 $t = substr( $t, 1 );
2166                                         } else {
2167                                                 // paragraph
2168                                                 if ( '' == trim($t) ) {
2169                                                         if ( $paragraphStack ) {
2170                                                                 $output .= $paragraphStack.'<br />';
2171                                                                 $paragraphStack = false;
2172                                                                 $this->mLastSection = 'p';
2173                                                         } else {
2174                                                                 if ($this->mLastSection != 'p' ) {
2175                                                                         $output .= $this->closeParagraph();
2176                                                                         $this->mLastSection = '';
2177                                                                         $paragraphStack = '<p>';
2178                                                                 } else {
2179                                                                         $paragraphStack = '</p><p>';
2180                                                                 }
2181                                                         }
2182                                                 } else {
2183                                                         if ( $paragraphStack ) {
2184                                                                 $output .= $paragraphStack;
2185                                                                 $paragraphStack = false;
2186                                                                 $this->mLastSection = 'p';
2187                                                         } else if ($this->mLastSection != 'p') {
2188                                                                 $output .= $this->closeParagraph().'<p>';
2189                                                                 $this->mLastSection = 'p';
2190                                                         }
2191                                                 }
2192                                         }
2193                                 }
2194                                 wfProfileOut( "$fname-paragraph" );
2195                         }
2196                         // somewhere above we forget to get out of pre block (bug 785)
2197                         if($preCloseMatch && $this->mInPre) {
2198                                 $this->mInPre = false;
2199                         }
2200                         if ($paragraphStack === false) {
2201                                 $output .= $t."\n";
2202                         }
2203                 }
2204                 while ( $prefixLength ) {
2205                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2206                         --$prefixLength;
2207                 }
2208                 if ( '' != $this->mLastSection ) {
2209                         $output .= '</' . $this->mLastSection . '>';
2210                         $this->mLastSection = '';
2211                 }
2212
2213                 wfProfileOut( $fname );
2214                 return $output;
2215         }
2216
2217         /**
2218          * Split up a string on ':', ignoring any occurences inside tags
2219          * to prevent illegal overlapping.
2220          * @param string $str the string to split
2221          * @param string &$before set to everything before the ':'
2222          * @param string &$after set to everything after the ':'
2223          * return string the position of the ':', or false if none found
2224          */
2225         function findColonNoLinks($str, &$before, &$after) {
2226                 $fname = 'Parser::findColonNoLinks';
2227                 wfProfileIn( $fname );
2228
2229                 $pos = strpos( $str, ':' );
2230                 if( $pos === false ) {
2231                         // Nothing to find!
2232                         wfProfileOut( $fname );
2233                         return false;
2234                 }
2235
2236                 $lt = strpos( $str, '<' );
2237                 if( $lt === false || $lt > $pos ) {
2238                         // Easy; no tag nesting to worry about
2239                         $before = substr( $str, 0, $pos );
2240                         $after = substr( $str, $pos+1 );
2241                         wfProfileOut( $fname );
2242                         return $pos;
2243                 }
2244
2245                 // Ugly state machine to walk through avoiding tags.
2246                 $state = self::COLON_STATE_TEXT;
2247                 $stack = 0;
2248                 $len = strlen( $str );
2249                 for( $i = 0; $i < $len; $i++ ) {
2250                         $c = $str{$i};
2251
2252                         switch( $state ) {
2253                         // (Using the number is a performance hack for common cases)
2254                         case 0: // self::COLON_STATE_TEXT:
2255                                 switch( $c ) {
2256                                 case "<":
2257                                         // Could be either a <start> tag or an </end> tag
2258                                         $state = self::COLON_STATE_TAGSTART;
2259                                         break;
2260                                 case ":":
2261                                         if( $stack == 0 ) {
2262                                                 // We found it!
2263                                                 $before = substr( $str, 0, $i );
2264                                                 $after = substr( $str, $i + 1 );
2265                                                 wfProfileOut( $fname );
2266                                                 return $i;
2267                                         }
2268                                         // Embedded in a tag; don't break it.
2269                                         break;
2270                                 default:
2271                                         // Skip ahead looking for something interesting
2272                                         $colon = strpos( $str, ':', $i );
2273                                         if( $colon === false ) {
2274                                                 // Nothing else interesting
2275                                                 wfProfileOut( $fname );
2276                                                 return false;
2277                                         }
2278                                         $lt = strpos( $str, '<', $i );
2279                                         if( $stack === 0 ) {
2280                                                 if( $lt === false || $colon < $lt ) {
2281                                                         // We found it!
2282                                                         $before = substr( $str, 0, $colon );
2283                                                         $after = substr( $str, $colon + 1 );
2284                                                         wfProfileOut( $fname );
2285                                                         return $i;
2286                                                 }
2287                                         }
2288                                         if( $lt === false ) {
2289                                                 // Nothing else interesting to find; abort!
2290                                                 // We're nested, but there's no close tags left. Abort!
2291                                                 break 2;
2292                                         }
2293                                         // Skip ahead to next tag start
2294                                         $i = $lt;
2295                                         $state = self::COLON_STATE_TAGSTART;
2296                                 }
2297                                 break;
2298                         case 1: // self::COLON_STATE_TAG:
2299                                 // In a <tag>
2300                                 switch( $c ) {
2301                                 case ">":
2302                                         $stack++;
2303                                         $state = self::COLON_STATE_TEXT;
2304                                         break;
2305                                 case "/":
2306                                         // Slash may be followed by >?
2307                                         $state = self::COLON_STATE_TAGSLASH;
2308                                         break;
2309                                 default:
2310                                         // ignore
2311                                 }
2312                                 break;
2313                         case 2: // self::COLON_STATE_TAGSTART:
2314                                 switch( $c ) {
2315                                 case "/":
2316                                         $state = self::COLON_STATE_CLOSETAG;
2317                                         break;
2318                                 case "!":
2319                                         $state = self::COLON_STATE_COMMENT;
2320                                         break;
2321                                 case ">":
2322                                         // Illegal early close? This shouldn't happen D:
2323                                         $state = self::COLON_STATE_TEXT;
2324                                         break;
2325                                 default:
2326                                         $state = self::COLON_STATE_TAG;
2327                                 }
2328                                 break;
2329                         case 3: // self::COLON_STATE_CLOSETAG:
2330                                 // In a </tag>
2331                                 if( $c == ">" ) {
2332                                         $stack--;
2333                                         if( $stack < 0 ) {
2334                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2335                                                 wfProfileOut( $fname );
2336                                                 return false;
2337                                         }
2338                                         $state = self::COLON_STATE_TEXT;
2339                                 }
2340                                 break;
2341                         case self::COLON_STATE_TAGSLASH:
2342                                 if( $c == ">" ) {
2343                                         // Yes, a self-closed tag <blah/>
2344                                         $state = self::COLON_STATE_TEXT;
2345                                 } else {
2346                                         // Probably we're jumping the gun, and this is an attribute
2347                                         $state = self::COLON_STATE_TAG;
2348                                 }
2349                                 break;
2350                         case 5: // self::COLON_STATE_COMMENT:
2351                                 if( $c == "-" ) {
2352                                         $state = self::COLON_STATE_COMMENTDASH;
2353                                 }
2354                                 break;
2355                         case self::COLON_STATE_COMMENTDASH:
2356                                 if( $c == "-" ) {
2357                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2358                                 } else {
2359                                         $state = self::COLON_STATE_COMMENT;
2360                                 }
2361                                 break;
2362                         case self::COLON_STATE_COMMENTDASHDASH:
2363                                 if( $c == ">" ) {
2364                                         $state = self::COLON_STATE_TEXT;
2365                                 } else {
2366                                         $state = self::COLON_STATE_COMMENT;
2367                                 }
2368                                 break;
2369                         default:
2370                                 throw new MWException( "State machine error in $fname" );
2371                         }
2372                 }
2373                 if( $stack > 0 ) {
2374                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2375                         return false;
2376                 }
2377                 wfProfileOut( $fname );
2378                 return false;
2379         }
2380
2381         /**
2382          * Return value of a magic variable (like PAGENAME)
2383          *
2384          * @private
2385          */
2386         function getVariableValue( $index ) {
2387                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2388
2389                 /**
2390                  * Some of these require message or data lookups and can be
2391                  * expensive to check many times.
2392                  */
2393                 static $varCache = array();
2394                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2395                         if ( isset( $varCache[$index] ) ) {
2396                                 return $varCache[$index];
2397                         }
2398                 }
2399
2400                 $ts = time();
2401                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2402
2403                 # Use the time zone
2404                 global $wgLocaltimezone;
2405                 if ( isset( $wgLocaltimezone ) ) {
2406                         $oldtz = getenv( 'TZ' );
2407                         putenv( 'TZ='.$wgLocaltimezone );
2408                 }
2409
2410                 wfSuppressWarnings(); // E_STRICT system time bitching
2411                 $localTimestamp = date( 'YmdHis', $ts );
2412                 $localMonth = date( 'm', $ts );
2413                 $localMonthName = date( 'n', $ts );
2414                 $localDay = date( 'j', $ts );
2415                 $localDay2 = date( 'd', $ts );
2416                 $localDayOfWeek = date( 'w', $ts );
2417                 $localWeek = date( 'W', $ts );
2418                 $localYear = date( 'Y', $ts );
2419                 $localHour = date( 'H', $ts );
2420                 if ( isset( $wgLocaltimezone ) ) {
2421                         putenv( 'TZ='.$oldtz );
2422                 }
2423                 wfRestoreWarnings();
2424
2425                 switch ( $index ) {
2426                         case 'currentmonth':
2427                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2428                         case 'currentmonthname':
2429                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2430                         case 'currentmonthnamegen':
2431                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2432                         case 'currentmonthabbrev':
2433                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2434                         case 'currentday':
2435                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2436                         case 'currentday2':
2437                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2438                         case 'localmonth':
2439                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2440                         case 'localmonthname':
2441                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2442                         case 'localmonthnamegen':
2443                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2444                         case 'localmonthabbrev':
2445                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2446                         case 'localday':
2447                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2448                         case 'localday2':
2449                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2450                         case 'pagename':
2451                                 return wfEscapeWikiText( $this->mTitle->getText() );
2452                         case 'pagenamee':
2453                                 return $this->mTitle->getPartialURL();
2454                         case 'fullpagename':
2455                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2456                         case 'fullpagenamee':
2457                                 return $this->mTitle->getPrefixedURL();
2458                         case 'subpagename':
2459                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2460                         case 'subpagenamee':
2461                                 return $this->mTitle->getSubpageUrlForm();
2462                         case 'basepagename':
2463                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2464                         case 'basepagenamee':
2465                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2466                         case 'talkpagename':
2467                                 if( $this->mTitle->canTalk() ) {
2468                                         $talkPage = $this->mTitle->getTalkPage();
2469                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2470                                 } else {
2471                                         return '';
2472                                 }
2473                         case 'talkpagenamee':
2474                                 if( $this->mTitle->canTalk() ) {
2475                                         $talkPage = $this->mTitle->getTalkPage();
2476                                         return $talkPage->getPrefixedUrl();
2477                                 } else {
2478                                         return '';
2479                                 }
2480                         case 'subjectpagename':
2481                                 $subjPage = $this->mTitle->getSubjectPage();
2482                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2483                         case 'subjectpagenamee':
2484                                 $subjPage = $this->mTitle->getSubjectPage();
2485                                 return $subjPage->getPrefixedUrl();
2486                         case 'revisionid':
2487                                 return $this->mRevisionId;
2488                         case 'revisionday':
2489                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2490                         case 'revisionday2':
2491                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2492                         case 'revisionmonth':
2493                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2494                         case 'revisionyear':
2495                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2496                         case 'revisiontimestamp':
2497                                 return $this->getRevisionTimestamp();
2498                         case 'namespace':
2499                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2500                         case 'namespacee':
2501                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2502                         case 'talkspace':
2503                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2504                         case 'talkspacee':
2505                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2506                         case 'subjectspace':
2507                                 return $this->mTitle->getSubjectNsText();
2508                         case 'subjectspacee':
2509                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2510                         case 'currentdayname':
2511                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2512                         case 'currentyear':
2513                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2514                         case 'currenttime':
2515                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2516                         case 'currenthour':
2517                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2518                         case 'currentweek':
2519                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2520                                 // int to remove the padding
2521                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2522                         case 'currentdow':
2523                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2524                         case 'localdayname':
2525                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2526                         case 'localyear':
2527                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2528                         case 'localtime':
2529                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2530                         case 'localhour':
2531                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2532                         case 'localweek':
2533                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2534                                 // int to remove the padding
2535                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2536                         case 'localdow':
2537                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2538                         case 'numberofarticles':
2539                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2540                         case 'numberoffiles':
2541                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2542                         case 'numberofusers':
2543                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2544                         case 'numberofpages':
2545                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2546                         case 'numberofadmins':
2547                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2548                         case 'numberofedits':
2549                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::edits() );
2550                         case 'currenttimestamp':
2551                                 return $varCache[$index] = wfTimestampNow();
2552                         case 'localtimestamp':
2553                                 return $varCache[$index] = $localTimestamp;
2554                         case 'currentversion':
2555                                 return $varCache[$index] = SpecialVersion::getVersion();
2556                         case 'sitename':
2557                                 return $wgSitename;
2558                         case 'server':
2559                                 return $wgServer;
2560                         case 'servername':
2561                                 return $wgServerName;
2562                         case 'scriptpath':
2563                                 return $wgScriptPath;
2564                         case 'directionmark':
2565                                 return $wgContLang->getDirMark();
2566                         case 'contentlanguage':
2567                                 global $wgContLanguageCode;
2568                                 return $wgContLanguageCode;
2569                         default:
2570                                 $ret = null;
2571                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2572                                         return $ret;
2573                                 else
2574                                         return null;
2575                 }
2576         }
2577
2578         /**
2579          * initialise the magic variables (like CURRENTMONTHNAME)
2580          *
2581          * @private
2582          */
2583         function initialiseVariables() {
2584                 $fname = 'Parser::initialiseVariables';
2585                 wfProfileIn( $fname );
2586                 $variableIDs = MagicWord::getVariableIDs();
2587
2588                 $this->mVariables = array();
2589                 foreach ( $variableIDs as $id ) {
2590                         $mw =& MagicWord::get( $id );
2591                         $mw->addToArray( $this->mVariables, $id );
2592                 }
2593                 wfProfileOut( $fname );
2594         }
2595
2596         /**
2597          * parse any parentheses in format ((title|part|part))
2598          * and call callbacks to get a replacement text for any found piece
2599          *
2600          * @param string $text The text to parse
2601          * @param array $callbacks rules in form:
2602          *     '{' => array(                            # opening parentheses
2603          *                                      'end' => '}',   # closing parentheses
2604          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2605          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2606          *                                                                )
2607          *                                      )
2608          *                                      'min' => 2,     # Minimum parenthesis count in cb
2609          *                                      'max' => 3,     # Maximum parenthesis count in cb
2610          * @private
2611          */
2612         function replace_callback ($text, $callbacks) {
2613                 wfProfileIn( __METHOD__ );
2614                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2615                 $lastOpeningBrace = -1;                 # last not closed parentheses
2616
2617                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2618
2619                 $i = 0;
2620                 while ( $i < strlen( $text ) ) {
2621                         # Find next opening brace, closing brace or pipe
2622                         if ( $lastOpeningBrace == -1 ) {
2623                                 $currentClosing = '';
2624                                 $search = $validOpeningBraces;
2625                         } else {
2626                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2627                                 $search = $validOpeningBraces . '|' . $currentClosing;
2628                         }
2629                         $rule = null;
2630                         $i += strcspn( $text, $search, $i );
2631                         if ( $i < strlen( $text ) ) {
2632                                 if ( $text[$i] == '|' ) {
2633                                         $found = 'pipe';
2634                                 } elseif ( $text[$i] == $currentClosing ) {
2635                                         $found = 'close';
2636                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2637                                         $found = 'open';
2638                                         $rule = $callbacks[$text[$i]];
2639                                 } else {
2640                                         # Some versions of PHP have a strcspn which stops on null characters
2641                                         # Ignore and continue
2642                                         ++$i;
2643                                         continue;
2644                                 }
2645                         } else {
2646                                 # All done
2647                                 break;
2648                         }
2649
2650                         if ( $found == 'open' ) {
2651                                 # found opening brace, let's add it to parentheses stack
2652                                 $piece = array('brace' => $text[$i],
2653                                                            'braceEnd' => $rule['end'],
2654                                                            'title' => '',
2655                                                            'parts' => null);
2656
2657                                 # count opening brace characters
2658                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2659                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2660                                 $i += $piece['count'];
2661
2662                                 # we need to add to stack only if opening brace count is enough for one of the rules
2663                                 if ( $piece['count'] >= $rule['min'] ) {
2664                                         $lastOpeningBrace ++;
2665                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2666                                 }
2667                         } elseif ( $found == 'close' ) {
2668                                 # lets check if it is enough characters for closing brace
2669                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2670                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2671
2672                                 # check for maximum matching characters (if there are 5 closing
2673                                 # characters, we will probably need only 3 - depending on the rules)
2674                                 $matchingCount = 0;
2675                                 $matchingCallback = null;
2676                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2677                                 if ( $count > $cbType['max'] ) {
2678                                         # The specified maximum exists in the callback array, unless the caller
2679                                         # has made an error
2680                                         $matchingCount = $cbType['max'];
2681                                 } else {
2682                                         # Count is less than the maximum
2683                                         # Skip any gaps in the callback array to find the true largest match
2684                                         # Need to use array_key_exists not isset because the callback can be null
2685                                         $matchingCount = $count;
2686                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2687                                                 --$matchingCount;
2688                                         }
2689                                 }
2690
2691                                 if ($matchingCount <= 0) {
2692                                         $i += $count;
2693                                         continue;
2694                                 }
2695                                 $matchingCallback = $cbType['cb'][$matchingCount];
2696
2697                                 # let's set a title or last part (if '|' was found)
2698                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2699                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2700                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2701                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2702                                 } else {
2703                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2704                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2705                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2706                                 }
2707
2708                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2709                                 $pieceEnd = $i + $matchingCount;
2710
2711                                 if( is_callable( $matchingCallback ) ) {
2712                                         $cbArgs = array (
2713                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2714                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2715                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2716                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2717                                                                          );
2718                                         # finally we can call a user callback and replace piece of text
2719                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2720                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2721                                         $i = $pieceStart + strlen($replaceWith);
2722                                 } else {
2723                                         # null value for callback means that parentheses should be parsed, but not replaced
2724                                         $i += $matchingCount;
2725                                 }
2726
2727                                 # reset last opening parentheses, but keep it in case there are unused characters
2728                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2729                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2730                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2731                                                            'title' => '',
2732                                                            'parts' => null,
2733                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2734                                 $openingBraceStack[$lastOpeningBrace--] = null;
2735
2736                                 if ($matchingCount < $piece['count']) {
2737                                         $piece['count'] -= $matchingCount;
2738                                         $piece['startAt'] -= $matchingCount;
2739                                         $piece['partStart'] = $piece['startAt'];
2740                                         # do we still qualify for any callback with remaining count?
2741                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2742                                         while ( $piece['count'] ) {
2743                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2744                                                         $lastOpeningBrace++;
2745                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2746                                                         break;
2747                                                 }
2748                                                 --$piece['count'];
2749                                         }
2750                                 }
2751                         } elseif ( $found == 'pipe' ) {
2752                                 # lets set a title if it is a first separator, or next part otherwise
2753                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2754                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2755                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2756                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2757                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2758                                 } else {
2759                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2760                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2761                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2762                                 }
2763                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2764                         }
2765                 }
2766
2767                 wfProfileOut( __METHOD__ );
2768                 return $text;
2769         }
2770
2771         /**
2772          * Replace magic variables, templates, and template arguments
2773          * with the appropriate text. Templates are substituted recursively,
2774          * taking care to avoid infinite loops.
2775          *
2776          * Note that the substitution depends on value of $mOutputType:
2777          *  self::OT_WIKI: only {{subst:}} templates
2778          *  self::OT_MSG: only magic variables
2779          *  self::OT_HTML: all templates and magic variables
2780          *
2781          * @param string $tex The text to transform
2782          * @param array $args Key-value pairs representing template parameters to substitute
2783          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2784          * @private
2785          */
2786         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2787                 # Prevent too big inclusions
2788                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2789                         return $text;
2790                 }
2791
2792                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2793                 wfProfileIn( $fname );
2794
2795                 # This function is called recursively. To keep track of arguments we need a stack:
2796                 array_push( $this->mArgStack, $args );
2797
2798                 $braceCallbacks = array();
2799                 if ( !$argsOnly ) {
2800                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2801                 }
2802                 if ( $this->mOutputType != self::OT_MSG ) {
2803                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2804                 }
2805                 if ( $braceCallbacks ) {
2806                         $callbacks = array(
2807                                 '{' => array(
2808                                         'end' => '}',
2809                                         'cb' => $braceCallbacks,
2810                                         'min' => $argsOnly ? 3 : 2,
2811                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2812                                 ),
2813                                 '[' => array(
2814                                         'end' => ']',
2815                                         'cb' => array(2=>null),
2816                                         'min' => 2,
2817                                         'max' => 2,
2818                                 )
2819                         );
2820                         $text = $this->replace_callback ($text, $callbacks);
2821
2822                         array_pop( $this->mArgStack );
2823                 }
2824                 wfProfileOut( $fname );
2825                 return $text;
2826         }
2827
2828         /**
2829          * Replace magic variables
2830          * @private
2831          */
2832         function variableSubstitution( $matches ) {
2833                 global $wgContLang;
2834                 $fname = 'Parser::variableSubstitution';
2835                 $varname = $wgContLang->lc($matches[1]);
2836                 wfProfileIn( $fname );
2837                 $skip = false;
2838                 if ( $this->mOutputType == self::OT_WIKI ) {
2839                         # Do only magic variables prefixed by SUBST
2840                         $mwSubst =& MagicWord::get( 'subst' );
2841                         if (!$mwSubst->matchStartAndRemove( $varname ))
2842                                 $skip = true;
2843                         # Note that if we don't substitute the variable below,
2844                         # we don't remove the {{subst:}} magic word, in case
2845                         # it is a template rather than a magic variable.
2846                 }
2847                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2848                         $id = $this->mVariables[$varname];
2849                         # Now check if we did really match, case sensitive or not
2850                         $mw =& MagicWord::get( $id );
2851                         if ($mw->match($matches[1])) {
2852                                 $text = $this->getVariableValue( $id );
2853                                 if (MagicWord::getCacheTTL($id)>-1)
2854                                         $this->mOutput->mContainsOldMagic = true;
2855                         } else {
2856                                 $text = $matches[0];
2857                         }
2858                 } else {
2859                         $text = $matches[0];
2860                 }
2861                 wfProfileOut( $fname );
2862                 return $text;
2863         }
2864
2865
2866         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2867         static function createAssocArgs( $args ) {
2868                 $assocArgs = array();
2869                 $index = 1;
2870                 foreach( $args as $arg ) {
2871                         $eqpos = strpos( $arg, '=' );
2872                         if ( $eqpos === false ) {
2873                                 $assocArgs[$index++] = $arg;
2874                         } else {
2875                                 $name = trim( substr( $arg, 0, $eqpos ) );
2876                                 $value = trim( substr( $arg, $eqpos+1 ) );
2877                                 if ( $value === false ) {
2878                                         $value = '';
2879                                 }
2880                                 if ( $name !== false ) {
2881                                         $assocArgs[$name] = $value;
2882                                 }
2883                         }
2884                 }
2885
2886                 return $assocArgs;
2887         }
2888
2889         /**
2890          * Return the text of a template, after recursively
2891          * replacing any variables or templates within the template.
2892          *
2893          * @param array $piece The parts of the template
2894          *  $piece['text']: matched text
2895          *  $piece['title']: the title, i.e. the part before the |
2896          *  $piece['parts']: the parameter array
2897          * @return string the text of the template
2898          * @private
2899          */
2900         function braceSubstitution( $piece ) {
2901                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2902                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2903                 wfProfileIn( $fname );
2904                 wfProfileIn( __METHOD__.'-setup' );
2905
2906                 # Flags
2907                 $found = false;             # $text has been filled
2908                 $nowiki = false;            # wiki markup in $text should be escaped
2909                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2910                 $noargs = false;            # Don't replace triple-brace arguments in $text
2911                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2912                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2913                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2914                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2915
2916                 # Title object, where $text came from
2917                 $title = NULL;
2918
2919                 $linestart = '';
2920
2921
2922                 # $part1 is the bit before the first |, and must contain only title characters
2923                 # $args is a list of arguments, starting from index 0, not including $part1
2924
2925                 $titleText = $part1 = $piece['title'];
2926                 # If the third subpattern matched anything, it will start with |
2927
2928                 if (null == $piece['parts']) {
2929                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2930                         if ($replaceWith != $piece['text']) {
2931                                 $text = $replaceWith;
2932                                 $found = true;
2933                                 $noparse = true;
2934                                 $noargs = true;
2935                         }
2936                 }
2937
2938                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2939                 wfProfileOut( __METHOD__.'-setup' );
2940
2941                 # SUBST
2942                 wfProfileIn( __METHOD__.'-modifiers' );
2943                 if ( !$found ) {
2944                         $mwSubst =& MagicWord::get( 'subst' );
2945                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2946                                 # One of two possibilities is true:
2947                                 # 1) Found SUBST but not in the PST phase
2948                                 # 2) Didn't find SUBST and in the PST phase
2949                                 # In either case, return without further processing
2950                                 $text = $piece['text'];
2951                                 $found = true;
2952                                 $noparse = true;
2953                                 $noargs = true;
2954                         }
2955                 }
2956
2957                 # MSG, MSGNW and RAW
2958                 if ( !$found ) {
2959                         # Check for MSGNW:
2960                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2961                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2962                                 $nowiki = true;
2963                         } else {
2964                                 # Remove obsolete MSG:
2965                                 $mwMsg =& MagicWord::get( 'msg' );
2966                                 $mwMsg->matchStartAndRemove( $part1 );
2967                         }
2968
2969                         # Check for RAW:
2970                         $mwRaw =& MagicWord::get( 'raw' );
2971                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2972                                 $forceRawInterwiki = true;
2973                         }
2974                 }
2975                 wfProfileOut( __METHOD__.'-modifiers' );
2976
2977                 //save path level before recursing into functions & templates.
2978                 $lastPathLevel = $this->mTemplatePath;
2979
2980                 # Parser functions
2981                 if ( !$found ) {
2982                         wfProfileIn( __METHOD__ . '-pfunc' );
2983
2984                         $colonPos = strpos( $part1, ':' );
2985                         if ( $colonPos !== false ) {
2986                                 # Case sensitive functions
2987                                 $function = substr( $part1, 0, $colonPos );
2988                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2989                                         $function = $this->mFunctionSynonyms[1][$function];
2990                                 } else {
2991                                         # Case insensitive functions
2992                                         $function = strtolower( $function );
2993                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2994                                                 $function = $this->mFunctionSynonyms[0][$function];
2995                                         } else {
2996                                                 $function = false;
2997                                         }
2998                                 }
2999                                 if ( $function ) {
3000                                         $funcArgs = array_map( 'trim', $args );
3001                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
3002                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
3003                                         $found = true;
3004
3005                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3006                                         //$noargs = true;
3007                                         //$noparse = true;
3008
3009                                         if ( is_array( $result ) ) {
3010                                                 if ( isset( $result[0] ) ) {
3011                                                         $text = $linestart . $result[0];
3012                                                         unset( $result[0] );
3013                                                 }
3014
3015                                                 // Extract flags into the local scope
3016                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3017                                                 extract( $result );
3018                                         } else {
3019                                                 $text = $linestart . $result;
3020                                         }
3021                                 }
3022                         }
3023                         wfProfileOut( __METHOD__ . '-pfunc' );
3024                 }
3025
3026                 # Template table test
3027
3028                 # Did we encounter this template already? If yes, it is in the cache
3029                 # and we need to check for loops.
3030                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3031                         $found = true;
3032
3033                         # Infinite loop test
3034                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3035                                 $noparse = true;
3036                                 $noargs = true;
3037                                 $found = true;
3038                                 $text = $linestart .
3039                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3040                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3041                         } else {
3042                                 # set $text to cached message.
3043                                 $text = $linestart . $this->mTemplates[$piece['title']];
3044                                 #treat title for cached page the same as others
3045                                 $ns = NS_TEMPLATE;
3046                                 $subpage = '';
3047                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3048                                 if ($subpage !== '') {
3049                                   $ns = $this->mTitle->getNamespace();
3050                                 }
3051                                 $title = Title::newFromText( $part1, $ns );
3052                                 //used by include size checking
3053                                 $titleText = $title->getPrefixedText();
3054                                 //used by edit section links
3055                                 $replaceHeadings = true;
3056
3057                         }
3058                 }
3059
3060                 # Load from database
3061                 if ( !$found ) {
3062                         wfProfileIn( __METHOD__ . '-loadtpl' );
3063                         $ns = NS_TEMPLATE;
3064                         # declaring $subpage directly in the function call
3065                         # does not work correctly with references and breaks
3066                         # {{/subpage}}-style inclusions
3067                         $subpage = '';
3068                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3069                         if ($subpage !== '') {
3070                                 $ns = $this->mTitle->getNamespace();
3071                         }
3072                         $title = Title::newFromText( $part1, $ns );
3073
3074
3075                         if ( !is_null( $title ) ) {
3076                                 $titleText = $title->getPrefixedText();
3077                                 # Check for language variants if the template is not found
3078                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3079                                         $wgContLang->findVariantLink($part1, $title);
3080                                 }
3081
3082                                 if ( !$title->isExternal() ) {
3083                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3084                                                 $text = SpecialPage::capturePath( $title );
3085                                                 if ( is_string( $text ) ) {
3086                                                         $found = true;
3087                                                         $noparse = true;
3088                                                         $noargs = true;
3089                                                         $isHTML = true;
3090                                                         $this->disableCache();
3091                                                 }
3092                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3093                                                 $found = false; //access denied
3094                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3095                                         } else {
3096                                                 list($articleContent,$title) = $this->fetchTemplateAndtitle( $title );
3097                                                 if ( $articleContent !== false ) {
3098                                                         $found = true;
3099                                                         $text = $articleContent;
3100                                                         $replaceHeadings = true;
3101                                                 }
3102                                         }
3103
3104                                         # If the title is valid but undisplayable, make a link to it
3105                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3106                                                 $text = "[[:$titleText]]";
3107                                                 $found = true;
3108                                         }
3109                                 } elseif ( $title->isTrans() ) {
3110                                         // Interwiki transclusion
3111                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3112                                                 $text = $this->interwikiTransclude( $title, 'render' );
3113                                                 $isHTML = true;
3114                                                 $noparse = true;
3115                                         } else {
3116                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3117                                                 $replaceHeadings = true;
3118                                         }
3119                                         $found = true;
3120                                 }
3121
3122                                 # Template cache array insertion
3123                                 # Use the original $piece['title'] not the mangled $part1, so that
3124                                 # modifiers such as RAW: produce separate cache entries
3125                                 if( $found ) {
3126                                         if( $isHTML ) {
3127                                                 // A special page; don't store it in the template cache.
3128                                         } else {
3129                                                 $this->mTemplates[$piece['title']] = $text;
3130                                         }
3131                                         $text = $linestart . $text;
3132                                 }
3133                         }
3134                         wfProfileOut( __METHOD__ . '-loadtpl' );
3135                 }
3136
3137                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3138                         # Error, oversize inclusion
3139                         $text = $linestart .
3140                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3141                         $noparse = true;
3142                         $noargs = true;
3143                 }
3144
3145                 # Recursive parsing, escaping and link table handling
3146                 # Only for HTML output
3147                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3148                         $text = wfEscapeWikiText( $text );
3149                 } elseif ( !$this->ot['msg'] && $found ) {
3150                         if ( $noargs ) {
3151                                 $assocArgs = array();
3152                         } else {
3153                                 # Clean up argument array
3154                                 $assocArgs = self::createAssocArgs($args);
3155                                 # Add a new element to the templace recursion path
3156                                 $this->mTemplatePath[$part1] = 1;
3157                         }
3158
3159                         if ( !$noparse ) {
3160                                 # If there are any <onlyinclude> tags, only include them
3161                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3162                                         $replacer = new OnlyIncludeReplacer;
3163                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3164                                                 array( &$replacer, 'replace' ), $text );
3165                                         $text = $replacer->output;
3166                                 }
3167                                 # Remove <noinclude> sections and <includeonly> tags
3168                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3169                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3170
3171                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3172                                         # Strip <nowiki>, <pre>, etc.
3173                                         $text = $this->strip( $text, $this->mStripState );
3174                                         if ( $this->ot['html'] ) {
3175                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3176                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3177                                                 $text = Sanitizer::removeHTMLcomments( $text );
3178                                         }
3179                                 }
3180                                 $text = $this->replaceVariables( $text, $assocArgs );
3181
3182                                 # If the template begins with a table or block-level
3183                                 # element, it should be treated as beginning a new line.
3184                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3185                                         $text = "\n" . $text;
3186                                 }
3187                         } elseif ( !$noargs ) {
3188                                 # $noparse and !$noargs
3189                                 # Just replace the arguments, not any double-brace items
3190                                 # This is used for rendered interwiki transclusion
3191                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3192                         }
3193                 }
3194                 # Prune lower levels off the recursion check path
3195                 $this->mTemplatePath = $lastPathLevel;
3196
3197                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3198                         # Error, oversize inclusion
3199                         $text = $linestart .
3200                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3201                         $noparse = true;
3202                         $noargs = true;
3203                 }
3204
3205                 if ( !$found ) {
3206                         wfProfileOut( $fname );
3207                         return $piece['text'];
3208                 } else {
3209                         wfProfileIn( __METHOD__ . '-placeholders' );
3210                         if ( $isHTML ) {
3211                                 # Replace raw HTML by a placeholder
3212                                 # Add a blank line preceding, to prevent it from mucking up
3213                                 # immediately preceding headings
3214                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3215                         } else {
3216                                 # replace ==section headers==
3217                                 # XXX this needs to go away once we have a better parser.
3218                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3219                                         if( !is_null( $title ) )
3220                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3221                                         else
3222                                                 $encodedname = base64_encode("");
3223                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3224                                                 PREG_SPLIT_DELIM_CAPTURE);
3225                                         $text = '';
3226                                         $nsec = $headingOffset;
3227
3228                                         for( $i = 0; $i < count($m); $i += 2 ) {
3229                                                 $text .= $m[$i];
3230                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3231                                                 $hl = $m[$i + 1];
3232                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3233                                                         $text .= $hl;
3234                                                         continue;
3235                                                 }
3236                                                 $m2 = array();
3237                                                 preg_match('/^(={1,6})(.*?)(={1,6}\s*?)$/m', $hl, $m2);
3238                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3239                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3240
3241                                                 $nsec++;
3242                                         }
3243                                 }
3244                         }
3245                         wfProfileOut( __METHOD__ . '-placeholders' );
3246                 }
3247
3248                 # Prune lower levels off the recursion check path
3249                 $this->mTemplatePath = $lastPathLevel;
3250
3251                 if ( !$found ) {
3252                         wfProfileOut( $fname );
3253                         return $piece['text'];
3254                 } else {
3255                         wfProfileOut( $fname );
3256                         return $text;
3257                 }
3258         }
3259
3260         /**
3261          * Fetch the unparsed text of a template and register a reference to it.
3262          */
3263         function fetchTemplateAndTitle( $title ) {
3264                 $templateCb = $this->mOptions->getTemplateCallback();
3265                 $stuff = call_user_func( $templateCb, $title, $this );
3266                 $text = $stuff['text'];
3267                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3268                 if ( isset( $stuff['deps'] ) ) {
3269                         foreach ( $stuff['deps'] as $dep ) {
3270                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3271                         }
3272                 }
3273                 return array($text,$finalTitle);
3274         }
3275
3276         function fetchTemplate( $title ) {
3277                 $rv = $this->fetchTemplateAndtitle($title);
3278                 return $rv[0];
3279         }
3280
3281         /**
3282          * Static function to get a template
3283          * Can be overridden via ParserOptions::setTemplateCallback().
3284          *
3285          * Returns an associative array:
3286          *    text          The unparsed template text
3287          *    finalTitle    (Optional) The title after following redirects
3288          *    deps          (Optional) An array of associative array dependencies:
3289          *                       title:    The dependency title, to be registered in templatelinks
3290          *                       page_id:  The page_id of the title
3291          *                       rev_id:   The revision ID loaded
3292          */
3293         static function statelessFetchTemplate( $title, $parser=false ) {
3294                 $text = $skip = false;
3295                 $finalTitle = $title;
3296                 $deps = array();
3297
3298                 // Loop to fetch the article, with up to 1 redirect
3299                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3300                         # Give extensions a chance to select the revision instead
3301                         $id = false; // Assume current
3302                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3303
3304                         if( $skip ) {
3305                                 $text = false;
3306                                 $deps[] = array(
3307                                         'title' => $title,
3308                                         'page_id' => $title->getArticleID(),
3309                                         'rev_id' => null );
3310                                 break;
3311                         }
3312                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3313                         $rev_id = $rev ? $rev->getId() : 0;
3314
3315                         $deps[] = array(
3316                                 'title' => $title,
3317                                 'page_id' => $title->getArticleID(),
3318                                 'rev_id' => $rev_id );
3319
3320                         if( $rev ) {
3321                                 $text = $rev->getText();
3322                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3323                                 global $wgLang;
3324                                 $message = $wgLang->lcfirst( $title->getText() );
3325                                 $text = wfMsgForContentNoTrans( $message );
3326                                 if( wfEmptyMsg( $message, $text ) ) {
3327                                         $text = false;
3328                                         break;
3329                                 }
3330                         } else {
3331                                 break;
3332                         }
3333                         if ( $text === false ) {
3334                                 break;
3335                         }
3336                         // Redirect?
3337                         $finalTitle = $title;
3338                         $title = Title::newFromRedirect( $text );
3339                 }
3340                 return array(
3341                         'text' => $text,
3342                         'finalTitle' => $finalTitle,
3343                         'deps' => $deps );
3344         }
3345
3346         /**
3347          * Transclude an interwiki link.
3348          */
3349         function interwikiTransclude( $title, $action ) {
3350                 global $wgEnableScaryTranscluding;
3351
3352                 if (!$wgEnableScaryTranscluding)
3353                         return wfMsg('scarytranscludedisabled');
3354
3355                 $url = $title->getFullUrl( "action=$action" );
3356
3357                 if (strlen($url) > 255)
3358                         return wfMsg('scarytranscludetoolong');
3359                 return $this->fetchScaryTemplateMaybeFromCache($url);
3360         }
3361
3362         function fetchScaryTemplateMaybeFromCache($url) {
3363                 global $wgTranscludeCacheExpiry;
3364                 $dbr = wfGetDB(DB_SLAVE);
3365                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3366                                 array('tc_url' => $url));
3367                 if ($obj) {
3368                         $time = $obj->tc_time;
3369                         $text = $obj->tc_contents;
3370                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3371                                 return $text;
3372                         }
3373                 }
3374
3375                 $text = Http::get($url);
3376                 if (!$text)
3377                         return wfMsg('scarytranscludefailed', $url);
3378
3379                 $dbw = wfGetDB(DB_MASTER);
3380                 $dbw->replace('transcache', array('tc_url'), array(
3381                         'tc_url' => $url,
3382                         'tc_time' => time(),
3383                         'tc_contents' => $text));
3384                 return $text;
3385         }
3386
3387
3388         /**
3389          * Triple brace replacement -- used for template arguments
3390          * @private
3391          */
3392         function argSubstitution( $matches ) {
3393                 $arg = trim( $matches['title'] );
3394                 $text = $matches['text'];
3395                 $inputArgs = end( $this->mArgStack );
3396
3397                 if ( array_key_exists( $arg, $inputArgs ) ) {
3398                         $text = $inputArgs[$arg];
3399                 } else if (($this->mOutputType == self::OT_HTML || $this->mOutputType == self::OT_PREPROCESS ) &&
3400                 null != $matches['parts'] && count($matches['parts']) > 0) {
3401                         $text = $matches['parts'][0];
3402                 }
3403                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3404                         $text = $matches['text'] .
3405                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3406                 }
3407
3408                 return $text;
3409         }
3410
3411         /**
3412          * Increment an include size counter
3413          *
3414          * @param string $type The type of expansion
3415          * @param integer $size The size of the text
3416          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3417          */
3418         function incrementIncludeSize( $type, $size ) {
3419                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3420                         return false;
3421                 } else {
3422                         $this->mIncludeSizes[$type] += $size;
3423                         return true;
3424                 }
3425         }
3426
3427         /**
3428          * Detect __NOGALLERY__ magic word and set a placeholder
3429          */
3430         function stripNoGallery( &$text ) {
3431                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3432                 # do not add TOC
3433                 $mw = MagicWord::get( 'nogallery' );
3434                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3435         }
3436
3437         /**
3438          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3439          * placeholder that will then be replaced by the real TOC in
3440          * ->formatHeadings, this works because at this points real
3441          * comments will have already been discarded by the sanitizer.
3442          *
3443          * Any additional __TOC__ magic words left over will be discarded
3444          * as there can only be one TOC on the page.
3445          */
3446         function stripToc( $text ) {
3447                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3448                 # do not add TOC
3449                 $mw = MagicWord::get( 'notoc' );
3450                 if( $mw->matchAndRemove( $text ) ) {
3451                         $this->mShowToc = false;
3452                 }
3453
3454                 $mw = MagicWord::get( 'toc' );
3455                 if( $mw->match( $text ) ) {
3456                         $this->mShowToc = true;
3457                         $this->mForceTocPosition = true;
3458
3459                         // Set a placeholder. At the end we'll fill it in with the TOC.
3460                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3461
3462                         // Only keep the first one.
3463                         $text = $mw->replace( '', $text );
3464                 }
3465                 return $text;
3466         }
3467
3468         /**
3469          * This function accomplishes several tasks:
3470          * 1) Auto-number headings if that option is enabled
3471          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3472          * 3) Add a Table of contents on the top for users who have enabled the option
3473          * 4) Auto-anchor headings
3474          *
3475          * It loops through all headlines, collects the necessary data, then splits up the
3476          * string and re-inserts the newly formatted headlines.
3477          *
3478          * @param string $text
3479          * @param boolean $isMain
3480          * @private
3481          */
3482         function formatHeadings( $text, $isMain=true ) {
3483                 global $wgMaxTocLevel, $wgContLang;
3484
3485                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3486                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3487                         $showEditLink = 0;
3488                 } else {
3489                         $showEditLink = $this->mOptions->getEditSection();
3490                 }
3491
3492                 # Inhibit editsection links if requested in the page
3493                 $esw =& MagicWord::get( 'noeditsection' );
3494                 if( $esw->matchAndRemove( $text ) ) {
3495                         $showEditLink = 0;
3496                 }
3497
3498                 # Get all headlines for numbering them and adding funky stuff like [edit]
3499                 # links - this is for later, but we need the number of headlines right now
3500                 $matches = array();
3501                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3502
3503                 # if there are fewer than 4 headlines in the article, do not show TOC
3504                 # unless it's been explicitly enabled.
3505                 $enoughToc = $this->mShowToc &&
3506                         (($numMatches >= 4) || $this->mForceTocPosition);
3507
3508                 # Allow user to stipulate that a page should have a "new section"
3509                 # link added via __NEWSECTIONLINK__
3510                 $mw =& MagicWord::get( 'newsectionlink' );
3511                 if( $mw->matchAndRemove( $text ) )
3512                         $this->mOutput->setNewSection( true );
3513
3514                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3515                 # override above conditions and always show TOC above first header
3516                 $mw =& MagicWord::get( 'forcetoc' );
3517                 if ($mw->matchAndRemove( $text ) ) {
3518                         $this->mShowToc = true;
3519                         $enoughToc = true;
3520                 }
3521
3522                 # We need this to perform operations on the HTML
3523                 $sk = $this->mOptions->getSkin();
3524
3525                 # headline counter
3526                 $headlineCount = 0;
3527                 $sectionCount = 0; # headlineCount excluding template sections
3528                 $numVisible = 0;
3529
3530                 # Ugh .. the TOC should have neat indentation levels which can be
3531                 # passed to the skin functions. These are determined here
3532                 $toc = '';
3533                 $full = '';
3534                 $head = array();
3535                 $sublevelCount = array();
3536                 $levelCount = array();
3537                 $toclevel = 0;
3538                 $level = 0;
3539                 $prevlevel = 0;
3540                 $toclevel = 0;
3541                 $prevtoclevel = 0;
3542                 $tocraw = array();
3543
3544                 foreach( $matches[3] as $headline ) {
3545                         $istemplate = 0;
3546                         $templatetitle = '';
3547                         $templatesection = 0;
3548                         $numbering = '';
3549                         $mat = array();
3550                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3551                                 $istemplate = 1;
3552                                 $templatetitle = base64_decode($mat[1]);
3553                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3554                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3555                         }
3556
3557                         if( $toclevel ) {
3558                                 $prevlevel = $level;
3559                                 $prevtoclevel = $toclevel;
3560                         }
3561                         $level = $matches[1][$headlineCount];
3562
3563                         if( $doNumberHeadings || $enoughToc ) {
3564
3565                                 if ( $level > $prevlevel ) {
3566                                         # Increase TOC level
3567                                         $toclevel++;
3568                                         $sublevelCount[$toclevel] = 0;
3569                                         if( $toclevel<$wgMaxTocLevel ) {
3570                                                 $prevtoclevel = $toclevel;
3571                                                 $toc .= $sk->tocIndent();
3572                                                 $numVisible++;
3573                                         }
3574                                 }
3575                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3576                                         # Decrease TOC level, find level to jump to
3577
3578                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3579                                                 # Can only go down to level 1
3580                                                 $toclevel = 1;
3581                                         } else {
3582                                                 for ($i = $toclevel; $i > 0; $i--) {
3583                                                         if ( $levelCount[$i] == $level ) {
3584                                                                 # Found last matching level
3585                                                                 $toclevel = $i;
3586                                                                 break;
3587                                                         }
3588                                                         elseif ( $levelCount[$i] < $level ) {
3589                                                                 # Found first matching level below current level
3590                                                                 $toclevel = $i + 1;
3591                                                                 break;
3592                                                         }
3593                                                 }
3594                                         }
3595                                         if( $toclevel<$wgMaxTocLevel ) {
3596                                                 if($prevtoclevel < $wgMaxTocLevel) {
3597                                                         # Unindent only if the previous toc level was shown :p
3598                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3599                                                 } else {
3600                                                         $toc .= $sk->tocLineEnd();
3601                                                 }
3602                                         }
3603                                 }
3604                                 else {
3605                                         # No change in level, end TOC line
3606                                         if( $toclevel<$wgMaxTocLevel ) {
3607                                                 $toc .= $sk->tocLineEnd();
3608                                         }
3609                                 }
3610
3611                                 $levelCount[$toclevel] = $level;
3612
3613                                 # count number of headlines for each level
3614                                 @$sublevelCount[$toclevel]++;
3615                                 $dot = 0;
3616                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3617                                         if( !empty( $sublevelCount[$i] ) ) {
3618                                                 if( $dot ) {
3619                                                         $numbering .= '.';
3620                                                 }
3621                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3622                                                 $dot = 1;
3623                                         }
3624                                 }
3625                         }
3626
3627                         # The canonized header is a version of the header text safe to use for links
3628                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3629                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3630
3631                         # Remove link placeholders by the link text.
3632                         #     <!--LINK number-->
3633                         # turns into
3634                         #     link text with suffix
3635                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3636                                                             "\$this->mLinkHolders['texts'][\$1]",
3637                                                             $canonized_headline );
3638                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3639                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3640                                                             $canonized_headline );
3641
3642                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3643                         $tocline = preg_replace(
3644                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3645                                 array( '',                          '<$1>'),
3646                                 $canonized_headline
3647                         );
3648                         $tocline = trim( $tocline );
3649
3650                         # For the anchor, strip out HTML-y stuff period
3651                         $canonized_headline = preg_replace( '/<.*?'.'>/', '', $canonized_headline );
3652                         $canonized_headline = trim( $canonized_headline );
3653
3654                         # Save headline for section edit hint before it's escaped
3655                         $headline_hint = $canonized_headline;
3656                         $canonized_headline = Sanitizer::escapeId( $canonized_headline );
3657                         $refers[$headlineCount] = $canonized_headline;
3658
3659                         # count how many in assoc. array so we can track dupes in anchors
3660                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3661                         $refcount[$headlineCount]=$refers[$canonized_headline];
3662
3663                         # Don't number the heading if it is the only one (looks silly)
3664                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3665                                 # the two are different if the line contains a link
3666                                 $headline=$numbering . ' ' . $headline;
3667                         }
3668
3669                         # Create the anchor for linking from the TOC to the section
3670                         $anchor = $canonized_headline;
3671                         if($refcount[$headlineCount] > 1 ) {
3672                                 $anchor .= '_' . $refcount[$headlineCount];
3673                         }
3674                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3675                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3676                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3677                         }
3678                         # give headline the correct <h#> tag
3679                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3680                                 if( $istemplate )
3681                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3682                                 else
3683                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3684                         } else {
3685                                 $editlink = '';
3686                         }
3687                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3688
3689                         $headlineCount++;
3690                         if( !$istemplate )
3691                                 $sectionCount++;
3692                 }
3693
3694                 $this->mOutput->setSections( $tocraw );
3695
3696                 # Never ever show TOC if no headers
3697                 if( $numVisible < 1 ) {
3698                         $enoughToc = false;
3699                 }
3700
3701                 if( $enoughToc ) {
3702                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3703                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3704                         }
3705                         $toc = $sk->tocList( $toc );
3706                 }
3707
3708                 # split up and insert constructed headlines
3709
3710                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3711                 $i = 0;
3712
3713                 foreach( $blocks as $block ) {
3714                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3715                                 # This is the [edit] link that appears for the top block of text when
3716                                 # section editing is enabled
3717
3718                                 # Disabled because it broke block formatting
3719                                 # For example, a bullet point in the top line
3720                                 # $full .= $sk->editSectionLink(0);
3721                         }
3722                         $full .= $block;
3723                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3724                                 # Top anchor now in skin
3725                                 $full = $full.$toc;
3726                         }
3727
3728                         if( !empty( $head[$i] ) ) {
3729                                 $full .= $head[$i];
3730                         }
3731                         $i++;
3732                 }
3733                 if( $this->mForceTocPosition ) {
3734                         return str_replace( '<!--MWTOC-->', $toc, $full );
3735                 } else {
3736                         return $full;
3737                 }
3738         }
3739
3740         /**
3741          * Transform wiki markup when saving a page by doing \r\n -> \n
3742          * conversion, substitting signatures, {{subst:}} templates, etc.
3743          *
3744          * @param string $text the text to transform
3745          * @param Title &$title the Title object for the current article
3746          * @param User &$user the User object describing the current user
3747          * @param ParserOptions $options parsing options
3748          * @param bool $clearState whether to clear the parser state first
3749          * @return string the altered wiki markup
3750          * @public
3751          */
3752         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3753                 $this->mOptions = $options;
3754                 $this->mTitle =& $title;
3755                 $this->setOutputType( self::OT_WIKI );
3756
3757                 if ( $clearState ) {
3758                         $this->clearState();
3759                 }
3760
3761                 $stripState = new StripState;
3762                 $pairs = array(
3763                         "\r\n" => "\n",
3764                 );
3765                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3766                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3767                 $text = $this->pstPass2( $text, $stripState, $user );
3768                 $text = $stripState->unstripBoth( $text );
3769                 return $text;
3770         }
3771
3772         /**
3773          * Pre-save transform helper function
3774          * @private
3775          */
3776         function pstPass2( $text, &$stripState, $user ) {
3777                 global $wgContLang, $wgLocaltimezone;
3778
3779                 /* Note: This is the timestamp saved as hardcoded wikitext to
3780                  * the database, we use $wgContLang here in order to give
3781                  * everyone the same signature and use the default one rather
3782                  * than the one selected in each user's preferences.
3783                  */
3784                 if ( isset( $wgLocaltimezone ) ) {
3785                         $oldtz = getenv( 'TZ' );
3786                         putenv( 'TZ='.$wgLocaltimezone );
3787                 }
3788                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3789                   ' (' . date( 'T' ) . ')';
3790                 if ( isset( $wgLocaltimezone ) ) {
3791                         putenv( 'TZ='.$oldtz );
3792                 }
3793
3794                 # Variable replacement
3795                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3796                 $text = $this->replaceVariables( $text );
3797
3798                 # Strip out <nowiki> etc. added via replaceVariables
3799                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3800
3801                 # Signatures
3802                 $sigText = $this->getUserSig( $user );
3803                 $text = strtr( $text, array(
3804                         '~~~~~' => $d,
3805                         '~~~~' => "$sigText $d",
3806                         '~~~' => $sigText
3807                 ) );
3808
3809                 # Context links: [[|name]] and [[name (context)|]]
3810                 #
3811                 global $wgLegalTitleChars;
3812                 $tc = "[$wgLegalTitleChars]";
3813                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3814
3815                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3816                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3817                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3818
3819                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3820                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3821                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3822
3823                 $t = $this->mTitle->getText();
3824                 $m = array();
3825                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3826                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3827                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3828                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3829                 } else {
3830                         # if there's no context, don't bother duplicating the title
3831                         $text = preg_replace( $p2, '[[\\1]]', $text );
3832                 }
3833
3834                 # Trim trailing whitespace
3835                 $text = rtrim( $text );
3836
3837                 return $text;
3838         }
3839
3840         /**
3841          * Fetch the user's signature text, if any, and normalize to
3842          * validated, ready-to-insert wikitext.
3843          *
3844          * @param User $user
3845          * @return string
3846          * @private
3847          */
3848         function getUserSig( &$user ) {
3849                 global $wgMaxSigChars;
3850
3851                 $username = $user->getName();
3852                 $nickname = $user->getOption( 'nickname' );
3853                 $nickname = $nickname === '' ? $username : $nickname;
3854
3855                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3856                         $nickname = $username;
3857                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3858                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3859                         # Sig. might contain markup; validate this
3860                         if( $this->validateSig( $nickname ) !== false ) {
3861                                 # Validated; clean up (if needed) and return it
3862                                 return $this->cleanSig( $nickname, true );
3863                         } else {
3864                                 # Failed to validate; fall back to the default
3865                                 $nickname = $username;
3866                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3867                         }
3868                 }
3869
3870                 // Make sure nickname doesnt get a sig in a sig
3871                 $nickname = $this->cleanSigInSig( $nickname );
3872
3873                 # If we're still here, make it a link to the user page
3874                 $userText = wfEscapeWikiText( $username );
3875                 $nickText = wfEscapeWikiText( $nickname );
3876                 if ( $user->isAnon() )  {
3877                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3878                 } else {
3879                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3880                 }
3881         }
3882
3883         /**
3884          * Check that the user's signature contains no bad XML
3885          *
3886          * @param string $text
3887          * @return mixed An expanded string, or false if invalid.
3888          */
3889         function validateSig( $text ) {
3890                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3891         }
3892
3893         /**
3894          * Clean up signature text
3895          *
3896          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3897          * 2) Substitute all transclusions
3898          *
3899          * @param string $text
3900          * @param $parsing Whether we're cleaning (preferences save) or parsing
3901          * @return string Signature text
3902          */
3903         function cleanSig( $text, $parsing = false ) {
3904                 global $wgTitle;
3905                 $this->startExternalParse( $this->mTitle, new ParserOptions(), $parsing ? self::OT_WIKI : self::OT_MSG );
3906
3907                 $substWord = MagicWord::get( 'subst' );
3908                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3909                 $substText = '{{' . $substWord->getSynonym( 0 );
3910
3911                 $text = preg_replace( $substRegex, $substText, $text );
3912                 $text = $this->cleanSigInSig( $text );
3913                 $text = $this->replaceVariables( $text );
3914
3915                 $this->clearState();
3916                 return $text;
3917         }
3918
3919         /**
3920          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3921          * @param string $text
3922          * @return string Signature text with /~{3,5}/ removed
3923          */
3924         function cleanSigInSig( $text ) {
3925                 $text = preg_replace( '/~{3,5}/', '', $text );
3926                 return $text;
3927         }
3928
3929         /**
3930          * Set up some variables which are usually set up in parse()
3931          * so that an external function can call some class members with confidence
3932          * @public
3933          */
3934         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3935                 $this->mTitle =& $title;
3936                 $this->mOptions = $options;
3937                 $this->setOutputType( $outputType );
3938                 if ( $clearState ) {
3939                         $this->clearState();
3940                 }
3941         }
3942
3943         /**
3944          * Transform a MediaWiki message by replacing magic variables.
3945          *
3946          * @param string $text the text to transform
3947          * @param ParserOptions $options  options
3948          * @return string the text with variables substituted
3949          * @public
3950          */
3951         function transformMsg( $text, $options ) {
3952                 global $wgTitle;
3953                 static $executing = false;
3954
3955                 $fname = "Parser::transformMsg";
3956
3957                 # Guard against infinite recursion
3958                 if ( $executing ) {
3959                         return $text;
3960                 }
3961                 $executing = true;
3962
3963                 wfProfileIn($fname);
3964
3965                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3966                         $this->mTitle = $wgTitle;
3967                 } else {
3968                         $this->mTitle = Title::newFromText('msg');
3969                 }
3970                 $this->mOptions = $options;
3971                 $this->setOutputType( self::OT_MSG );
3972                 $this->clearState();
3973                 $text = $this->replaceVariables( $text );
3974
3975                 $executing = false;
3976                 wfProfileOut($fname);
3977                 return $text;
3978         }
3979
3980         /**
3981          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3982          * The callback should have the following form:
3983          *    function myParserHook( $text, $params, &$parser ) { ... }
3984          *
3985          * Transform and return $text. Use $parser for any required context, e.g. use
3986          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3987          *
3988          * @public
3989          *
3990          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3991          * @param mixed $callback The callback function (and object) to use for the tag
3992          *
3993          * @return The old value of the mTagHooks array associated with the hook
3994          */
3995         function setHook( $tag, $callback ) {
3996                 $tag = strtolower( $tag );
3997                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3998                 $this->mTagHooks[$tag] = $callback;
3999
4000                 return $oldVal;
4001         }
4002
4003         function setTransparentTagHook( $tag, $callback ) {
4004                 $tag = strtolower( $tag );
4005                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4006                 $this->mTransparentTagHooks[$tag] = $callback;
4007
4008                 return $oldVal;
4009         }
4010
4011         /**
4012          * Create a function, e.g. {{sum:1|2|3}}
4013          * The callback function should have the form:
4014          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4015          *
4016          * The callback may either return the text result of the function, or an array with the text
4017          * in element 0, and a number of flags in the other elements. The names of the flags are
4018          * specified in the keys. Valid flags are:
4019          *   found                     The text returned is valid, stop processing the template. This
4020          *                             is on by default.
4021          *   nowiki                    Wiki markup in the return value should be escaped
4022          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4023          *   noargs                    Don't replace triple-brace arguments in the return value
4024          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4025          *
4026          * @public
4027          *
4028          * @param string $id The magic word ID
4029          * @param mixed $callback The callback function (and object) to use
4030          * @param integer $flags a combination of the following flags:
4031          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4032          *
4033          * @return The old callback function for this name, if any
4034          */
4035         function setFunctionHook( $id, $callback, $flags = 0 ) {
4036                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
4037                 $this->mFunctionHooks[$id] = $callback;
4038
4039                 # Add to function cache
4040                 $mw = MagicWord::get( $id );
4041                 if( !$mw )
4042                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4043
4044                 $synonyms = $mw->getSynonyms();
4045                 $sensitive = intval( $mw->isCaseSensitive() );
4046
4047                 foreach ( $synonyms as $syn ) {
4048                         # Case
4049                         if ( !$sensitive ) {
4050                                 $syn = strtolower( $syn );
4051                         }
4052                         # Add leading hash
4053                         if ( !( $flags & SFH_NO_HASH ) ) {
4054                                 $syn = '#' . $syn;
4055                         }
4056                         # Remove trailing colon
4057                         if ( substr( $syn, -1, 1 ) == ':' ) {
4058                                 $syn = substr( $syn, 0, -1 );
4059                         }
4060                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4061                 }
4062                 return $oldVal;
4063         }
4064
4065         /**
4066          * Get all registered function hook identifiers
4067          *
4068          * @return array
4069          */
4070         function getFunctionHooks() {
4071                 return array_keys( $this->mFunctionHooks );
4072         }
4073
4074         /**
4075          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4076          * Placeholders created in Skin::makeLinkObj()
4077          * Returns an array of links found, indexed by PDBK:
4078          *  0 - broken
4079          *  1 - normal link
4080          *  2 - stub
4081          * $options is a bit field, RLH_FOR_UPDATE to select for update
4082          */
4083         function replaceLinkHolders( &$text, $options = 0 ) {
4084                 global $wgUser;
4085                 global $wgContLang;
4086
4087                 $fname = 'Parser::replaceLinkHolders';
4088                 wfProfileIn( $fname );
4089
4090                 $pdbks = array();
4091                 $colours = array();
4092                 $sk = $this->mOptions->getSkin();
4093                 $linkCache = LinkCache::singleton();
4094
4095                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4096                         wfProfileIn( $fname.'-check' );
4097                         $dbr = wfGetDB( DB_SLAVE );
4098                         $page = $dbr->tableName( 'page' );
4099                         $threshold = $wgUser->getOption('stubthreshold');
4100
4101                         # Sort by namespace
4102                         asort( $this->mLinkHolders['namespaces'] );
4103
4104                         # Generate query
4105                         $query = false;
4106                         $current = null;
4107                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4108                                 # Make title object
4109                                 $title = $this->mLinkHolders['titles'][$key];
4110
4111                                 # Skip invalid entries.
4112                                 # Result will be ugly, but prevents crash.
4113                                 if ( is_null( $title ) ) {
4114                                         continue;
4115                                 }
4116                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4117
4118                                 # Check if it's a static known link, e.g. interwiki
4119                                 if ( $title->isAlwaysKnown() ) {
4120                                         $colours[$pdbk] = 1;
4121                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4122                                         $colours[$pdbk] = 1;
4123                                         $this->mOutput->addLink( $title, $id );
4124                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4125                                         $colours[$pdbk] = 0;
4126                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4127                                         $colours[$pdbk] = 0;
4128                                 } else {
4129                                         # Not in the link cache, add it to the query
4130                                         if ( !isset( $current ) ) {
4131                                                 $current = $ns;
4132                                                 $query =  "SELECT page_id, page_namespace, page_title, page_len, page_is_redirect";
4133                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4134                                         } elseif ( $current != $ns ) {
4135                                                 $current = $ns;
4136                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4137                                         } else {
4138                                                 $query .= ', ';
4139                                         }
4140
4141                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4142                                 }
4143                         }
4144                         if ( $query ) {
4145                                 $query .= '))';
4146                                 if ( $options & RLH_FOR_UPDATE ) {
4147                                         $query .= ' FOR UPDATE';
4148                                 }
4149
4150                                 $res = $dbr->query( $query, $fname );
4151
4152                                 # Fetch data and form into an associative array
4153                                 # non-existent = broken
4154                                 # 1 = known
4155                                 # 2 = stub
4156                                 while ( $s = $dbr->fetchObject($res) ) {
4157                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4158                                         $pdbk = $title->getPrefixedDBkey();
4159                                         $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
4160                                         $this->mOutput->addLink( $title, $s->page_id );
4161
4162                                         $colours[$pdbk] = ( $threshold == 0 || (
4163                                                                 $s->page_len >= $threshold || # always true if $threshold <= 0
4164                                                                 $s->page_is_redirect ||
4165                                                                 !MWNamespace::isContent( $s->page_namespace ) )
4166                                                             ? 1 : 2 );
4167                                 }
4168                         }
4169                         wfProfileOut( $fname.'-check' );
4170
4171                         # Do a second query for different language variants of links and categories
4172                         if( $wgContLang->hasVariants() ) {
4173                                 $linkBatch = new LinkBatch();
4174                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4175                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4176                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4177
4178                                 $categories = $this->mOutput->getCategoryLinks();
4179
4180                                 // Add variants of links to link batch
4181                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4182                                         $title = $this->mLinkHolders['titles'][$key];
4183                                         if ( is_null( $title ) )
4184                                                 continue;
4185
4186                                         $pdbk = $title->getPrefixedDBkey();
4187                                         $titleText = $title->getText();
4188
4189                                         // generate all variants of the link title text
4190                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4191
4192                                         // if link was not found (in first query), add all variants to query
4193                                         if ( !isset($colours[$pdbk]) ){
4194                                                 foreach($allTextVariants as $textVariant){
4195                                                         if($textVariant != $titleText){
4196                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4197                                                                 if(is_null($variantTitle)) continue;
4198                                                                 $linkBatch->addObj( $variantTitle );
4199                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4200                                                         }
4201                                                 }
4202                                         }
4203                                 }
4204
4205                                 // process categories, check if a category exists in some variant
4206                                 foreach( $categories as $category ){
4207                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4208                                         foreach($variants as $variant){
4209                                                 if($variant != $category){
4210                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4211                                                         if(is_null($variantTitle)) continue;
4212                                                         $linkBatch->addObj( $variantTitle );
4213                                                         $categoryMap[$variant] = $category;
4214                                                 }
4215                                         }
4216                                 }
4217
4218
4219                                 if ( !$linkBatch->isEmpty() ){
4220                                         // construct query
4221                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4222
4223                                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_len, page_is_redirect";
4224
4225                                         $variantQuery .= " FROM $page WHERE $titleClause";
4226                                         if ( $options & RLH_FOR_UPDATE ) {
4227                                                 $variantQuery .= ' FOR UPDATE';
4228                                         }
4229
4230                                         $varRes = $dbr->query( $variantQuery, $fname );
4231
4232                                         // for each found variants, figure out link holders and replace
4233                                         while ( $s = $dbr->fetchObject($varRes) ) {
4234
4235                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4236                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4237                                                 $vardbk = $variantTitle->getDBkey();
4238
4239                                                 $holderKeys = array();
4240                                                 if(isset($variantMap[$varPdbk])){
4241                                                         $holderKeys = $variantMap[$varPdbk];
4242                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
4243                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4244                                                 }
4245
4246                                                 // loop over link holders
4247                                                 foreach($holderKeys as $key){
4248                                                         $title = $this->mLinkHolders['titles'][$key];
4249                                                         if ( is_null( $title ) ) continue;
4250
4251                                                         $pdbk = $title->getPrefixedDBkey();
4252
4253                                                         if(!isset($colours[$pdbk])){
4254                                                                 // found link in some of the variants, replace the link holder data
4255                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4256                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4257
4258                                                                 // set pdbk and colour
4259                                                                 $pdbks[$key] = $varPdbk;
4260                                                                 if ( $threshold >  0 ) {
4261                                                                         $size = $s->page_len;
4262                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4263                                                                                 $colours[$varPdbk] = 1;
4264                                                                         } else {
4265                                                                                 $colours[$varPdbk] = 2;
4266                                                                         }
4267                                                                 }
4268                                                                 else {
4269                                                                         $colours[$varPdbk] = 1;
4270                                                                 }
4271                                                         }
4272                                                 }
4273
4274                                                 // check if the object is a variant of a category
4275                                                 if(isset($categoryMap[$vardbk])){
4276                                                         $oldkey = $categoryMap[$vardbk];
4277                                                         if($oldkey != $vardbk)
4278                                                                 $varCategories[$oldkey]=$vardbk;
4279                                                 }
4280                                         }
4281
4282                                         // rebuild the categories in original order (if there are replacements)
4283                                         if(count($varCategories)>0){
4284                                                 $newCats = array();
4285                                                 $originalCats = $this->mOutput->getCategories();
4286                                                 foreach($originalCats as $cat => $sortkey){
4287                                                         // make the replacement
4288                                                         if( array_key_exists($cat,$varCategories) )
4289                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4290                                                         else $newCats[$cat] = $sortkey;
4291                                                 }
4292                                                 $this->mOutput->setCategoryLinks($newCats);
4293                                         }
4294                                 }
4295                         }
4296
4297                         # Construct search and replace arrays
4298                         wfProfileIn( $fname.'-construct' );
4299                         $replacePairs = array();
4300                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4301                                 $pdbk = $pdbks[$key];
4302                                 $searchkey = "<!--LINK $key-->";
4303                                 $title = $this->mLinkHolders['titles'][$key];
4304                                 if ( empty( $colours[$pdbk] ) ) {
4305                                         $linkCache->addBadLinkObj( $title );
4306                                         $colours[$pdbk] = 0;
4307                                         $this->mOutput->addLink( $title, 0 );
4308                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4309                                                                         $this->mLinkHolders['texts'][$key],
4310                                                                         $this->mLinkHolders['queries'][$key] );
4311                                 } elseif ( $colours[$pdbk] == 1 ) {
4312                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4313                                                                         $this->mLinkHolders['texts'][$key],
4314                                                                         $this->mLinkHolders['queries'][$key] );
4315                                 } elseif ( $colours[$pdbk] == 2 ) {
4316                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4317                                                                         $this->mLinkHolders['texts'][$key],
4318                                                                         $this->mLinkHolders['queries'][$key] );
4319                                 }
4320                         }
4321                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4322                         wfProfileOut( $fname.'-construct' );
4323
4324                         # Do the thing
4325                         wfProfileIn( $fname.'-replace' );
4326                         $text = preg_replace_callback(
4327                                 '/(<!--LINK .*?-->)/',
4328                                 $replacer->cb(),
4329                                 $text);
4330
4331                         wfProfileOut( $fname.'-replace' );
4332                 }
4333
4334                 # Now process interwiki link holders
4335                 # This is quite a bit simpler than internal links
4336                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4337                         wfProfileIn( $fname.'-interwiki' );
4338                         # Make interwiki link HTML
4339                         $replacePairs = array();
4340                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4341                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4342                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4343                         }
4344                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4345
4346                         $text = preg_replace_callback(
4347                                 '/<!--IWLINK (.*?)-->/',
4348                                 $replacer->cb(),
4349                                 $text );
4350                         wfProfileOut( $fname.'-interwiki' );
4351                 }
4352
4353                 wfProfileOut( $fname );
4354                 return $colours;
4355         }
4356
4357         /**
4358          * Replace <!--LINK--> link placeholders with plain text of links
4359          * (not HTML-formatted).
4360          * @param string $text
4361          * @return string
4362          */
4363         function replaceLinkHoldersText( $text ) {
4364                 $fname = 'Parser::replaceLinkHoldersText';
4365                 wfProfileIn( $fname );
4366
4367                 $text = preg_replace_callback(
4368                         '/<!--(LINK|IWLINK) (.*?)-->/',
4369                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4370                         $text );
4371
4372                 wfProfileOut( $fname );
4373                 return $text;
4374         }
4375
4376         /**
4377          * @param array $matches
4378          * @return string
4379          * @private
4380          */
4381         function replaceLinkHoldersTextCallback( $matches ) {
4382                 $type = $matches[1];
4383                 $key  = $matches[2];
4384                 if( $type == 'LINK' ) {
4385                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4386                                 return $this->mLinkHolders['texts'][$key];
4387                         }
4388                 } elseif( $type == 'IWLINK' ) {
4389                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4390                                 return $this->mInterwikiLinkHolders['texts'][$key];
4391                         }
4392                 }
4393                 return $matches[0];
4394         }
4395
4396         /**
4397          * Tag hook handler for 'pre'.
4398          */
4399         function renderPreTag( $text, $attribs ) {
4400                 // Backwards-compatibility hack
4401                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4402
4403                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4404                 return wfOpenElement( 'pre', $attribs ) .
4405                         Xml::escapeTagsOnly( $content ) .
4406                         '</pre>';
4407         }
4408
4409         /**
4410          * Renders an image gallery from a text with one line per image.
4411          * text labels may be given by using |-style alternative text. E.g.
4412          *   Image:one.jpg|The number "1"
4413          *   Image:tree.jpg|A tree
4414          * given as text will return the HTML of a gallery with two images,
4415          * labeled 'The number "1"' and
4416          * 'A tree'.
4417          */
4418         function renderImageGallery( $text, $params ) {
4419                 $ig = new ImageGallery();
4420                 $ig->setContextTitle( $this->mTitle );
4421                 $ig->setShowBytes( false );
4422                 $ig->setShowFilename( false );
4423                 $ig->setParser( $this );
4424                 $ig->setHideBadImages();
4425                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4426                 $ig->useSkin( $this->mOptions->getSkin() );
4427                 $ig->mRevisionId = $this->mRevisionId;
4428
4429                 if( isset( $params['caption'] ) ) {
4430                         $caption = $params['caption'];
4431                         $caption = htmlspecialchars( $caption );
4432                         $caption = $this->replaceInternalLinks( $caption );
4433                         $ig->setCaptionHtml( $caption );
4434                 }
4435                 if( isset( $params['perrow'] ) ) {
4436                         $ig->setPerRow( $params['perrow'] );
4437                 }
4438                 if( isset( $params['widths'] ) ) {
4439                         $ig->setWidths( $params['widths'] );
4440                 }
4441                 if( isset( $params['heights'] ) ) {
4442                         $ig->setHeights( $params['heights'] );
4443                 }
4444
4445                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4446
4447                 $lines = explode( "\n", $text );
4448                 foreach ( $lines as $line ) {
4449                         # match lines like these:
4450                         # Image:someimage.jpg|This is some image
4451                         $matches = array();
4452                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4453                         # Skip empty lines
4454                         if ( count( $matches ) == 0 ) {
4455                                 continue;
4456                         }
4457                         $tp = Title::newFromText( $matches[1] );
4458                         $nt =& $tp;
4459                         if( is_null( $nt ) ) {
4460                                 # Bogus title. Ignore these so we don't bomb out later.
4461                                 continue;
4462                         }
4463                         if ( isset( $matches[3] ) ) {
4464                                 $label = $matches[3];
4465                         } else {
4466                                 $label = '';
4467                         }
4468
4469                         $pout = $this->parse( $label,
4470                                 $this->mTitle,
4471                                 $this->mOptions,
4472                                 false, // Strip whitespace...?
4473                                 false  // Don't clear state!
4474                         );
4475                         $html = $pout->getText();
4476
4477                         $ig->add( $nt, $html );
4478
4479                         # Only add real images (bug #5586)
4480                         if ( $nt->getNamespace() == NS_IMAGE ) {
4481                                 $this->mOutput->addImage( $nt->getDBkey() );
4482                         }
4483                 }
4484                 return $ig->toHTML();
4485         }
4486
4487         function getImageParams( $handler ) {
4488                 if ( $handler ) {
4489                         $handlerClass = get_class( $handler );
4490                 } else {
4491                         $handlerClass = '';
4492                 }
4493                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4494                         // Initialise static lists
4495                         static $internalParamNames = array(
4496                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4497                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4498                                         'bottom', 'text-bottom' ),
4499                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4500                                         'upright', 'border' ),
4501                         );
4502                         static $internalParamMap;
4503                         if ( !$internalParamMap ) {
4504                                 $internalParamMap = array();
4505                                 foreach ( $internalParamNames as $type => $names ) {
4506                                         foreach ( $names as $name ) {
4507                                                 $magicName = str_replace( '-', '_', "img_$name" );
4508                                                 $internalParamMap[$magicName] = array( $type, $name );
4509                                         }
4510                                 }
4511                         }
4512
4513                         // Add handler params
4514                         $paramMap = $internalParamMap;
4515                         if ( $handler ) {
4516                                 $handlerParamMap = $handler->getParamMap();
4517                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4518                                         $paramMap[$magic] = array( 'handler', $paramName );
4519                                 }
4520                         }
4521                         $this->mImageParams[$handlerClass] = $paramMap;
4522                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4523                 }
4524                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4525         }
4526
4527         /**
4528          * Parse image options text and use it to make an image
4529          */
4530         function makeImage( $title, $options ) {
4531                 # @TODO: let the MediaHandler specify its transform parameters
4532                 #
4533                 # Check if the options text is of the form "options|alt text"
4534                 # Options are:
4535                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4536                 #  * left               no resizing, just left align. label is used for alt= only
4537                 #  * right              same, but right aligned
4538                 #  * none               same, but not aligned
4539                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4540                 #  * center             center the image
4541                 #  * framed             Keep original image size, no magnify-button.
4542                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4543                 #  * upright            reduce width for upright images, rounded to full __0 px
4544                 #  * border             draw a 1px border around the image
4545                 # vertical-align values (no % or length right now):
4546                 #  * baseline
4547                 #  * sub
4548                 #  * super
4549                 #  * top
4550                 #  * text-top
4551                 #  * middle
4552                 #  * bottom
4553                 #  * text-bottom
4554
4555                 $parts = array_map( 'trim', explode( '|', $options) );
4556                 $sk = $this->mOptions->getSkin();
4557
4558                 # Give extensions a chance to select the file revision for us
4559                 $skip = $time = false;
4560                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4561
4562                 if ( $skip ) {
4563                         return $sk->makeLinkObj( $title );
4564                 }
4565
4566                 # Get parameter map
4567                 $file = wfFindFile( $title, $time );
4568                 $handler = $file ? $file->getHandler() : false;
4569
4570                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4571
4572                 # Process the input parameters
4573                 $caption = '';
4574                 $params = array( 'frame' => array(), 'handler' => array(),
4575                         'horizAlign' => array(), 'vertAlign' => array() );
4576                 foreach( $parts as $part ) {
4577                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4578                         if ( isset( $paramMap[$magicName] ) ) {
4579                                 list( $type, $paramName ) = $paramMap[$magicName];
4580                                 $params[$type][$paramName] = $value;
4581
4582                                 // Special case; width and height come in one variable together
4583                                 if( $type == 'handler' && $paramName == 'width' ) {
4584                                         $m = array();
4585                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4586                                                 $params[$type]['width'] = intval( $m[1] );
4587                                                 $params[$type]['height'] = intval( $m[2] );
4588                                         } else {
4589                                                 $params[$type]['width'] = intval( $value );
4590                                         }
4591                                 }
4592                         } else {
4593                                 $caption = $part;
4594                         }
4595                 }
4596
4597                 # Process alignment parameters
4598                 if ( $params['horizAlign'] ) {
4599                         $params['frame']['align'] = key( $params['horizAlign'] );
4600                 }
4601                 if ( $params['vertAlign'] ) {
4602                         $params['frame']['valign'] = key( $params['vertAlign'] );
4603                 }
4604
4605                 # Validate the handler parameters
4606                 if ( $handler ) {
4607                         foreach ( $params['handler'] as $name => $value ) {
4608                                 if ( !$handler->validateParam( $name, $value ) ) {
4609                                         unset( $params['handler'][$name] );
4610                                 }
4611                         }
4612                 }
4613
4614                 # Strip bad stuff out of the alt text
4615                 $alt = $this->replaceLinkHoldersText( $caption );
4616
4617                 # make sure there are no placeholders in thumbnail attributes
4618                 # that are later expanded to html- so expand them now and
4619                 # remove the tags
4620                 $alt = $this->mStripState->unstripBoth( $alt );
4621                 $alt = Sanitizer::stripAllTags( $alt );
4622
4623                 $params['frame']['alt'] = $alt;
4624                 $params['frame']['caption'] = $caption;
4625
4626                 # Linker does the rest
4627                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4628
4629                 # Give the handler a chance to modify the parser object
4630                 if ( $handler ) {
4631                         $handler->parserTransformHook( $this, $file );
4632                 }
4633
4634                 return $ret;
4635         }
4636
4637         /**
4638          * Set a flag in the output object indicating that the content is dynamic and
4639          * shouldn't be cached.
4640          */
4641         function disableCache() {
4642                 wfDebug( "Parser output marked as uncacheable.\n" );
4643                 $this->mOutput->mCacheTime = -1;
4644         }
4645
4646         /**#@+
4647          * Callback from the Sanitizer for expanding items found in HTML attribute
4648          * values, so they can be safely tested and escaped.
4649          * @param string $text
4650          * @param array $args
4651          * @return string
4652          * @private
4653          */
4654         function attributeStripCallback( &$text, $args ) {
4655                 $text = $this->replaceVariables( $text, $args );
4656                 $text = $this->mStripState->unstripBoth( $text );
4657                 return $text;
4658         }
4659
4660         /**#@-*/
4661
4662         /**#@+
4663          * Accessor/mutator
4664          */
4665         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4666         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4667         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4668         /**#@-*/
4669
4670         /**#@+
4671          * Accessor
4672          */
4673         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4674         /**#@-*/
4675
4676
4677         /**
4678          * Break wikitext input into sections, and either pull or replace
4679          * some particular section's text.
4680          *
4681          * External callers should use the getSection and replaceSection methods.
4682          *
4683          * @param $text Page wikitext
4684          * @param $section Numbered section. 0 pulls the text before the first
4685          *                 heading; other numbers will pull the given section
4686          *                 along with its lower-level subsections.
4687          * @param $mode One of "get" or "replace"
4688          * @param $newtext Replacement text for section data.
4689          * @return string for "get", the extracted section text.
4690          *                for "replace", the whole page with the section replaced.
4691          */
4692         private function extractSections( $text, $section, $mode, $newtext='' ) {
4693                 # I.... _hope_ this is right.
4694                 # Otherwise, sometimes we don't have things initialized properly.
4695                 $this->clearState();
4696
4697                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4698                 # comments to be stripped as well)
4699                 $stripState = new StripState;
4700
4701                 $oldOutputType = $this->mOutputType;
4702                 $oldOptions = $this->mOptions;
4703                 $this->mOptions = new ParserOptions();
4704                 $this->setOutputType( self::OT_WIKI );
4705
4706                 $striptext = $this->strip( $text, $stripState, true );
4707
4708                 $this->setOutputType( $oldOutputType );
4709                 $this->mOptions = $oldOptions;
4710
4711                 # now that we can be sure that no pseudo-sections are in the source,
4712                 # split it up by section
4713                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4714                 $comment = "(?:$uniq-!--.*?QINU\x07)";
4715                 $secs = preg_split(
4716                         "/
4717                         (
4718                                 ^
4719                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4720                                 (=+) # Should this be limited to 6?
4721                                 .+?  # Section title...
4722                                 \\2  # Ending = count must match start
4723                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4724                                 $
4725                         |
4726                                 <h([1-6])\b.*?>
4727                                 .*?
4728                                 <\/h\\3\s*>
4729                         )
4730                         /mix",
4731                         $striptext, -1,
4732                         PREG_SPLIT_DELIM_CAPTURE);
4733
4734                 if( $mode == "get" ) {
4735                         if( $section == 0 ) {
4736                                 // "Section 0" returns the content before any other section.
4737                                 $rv = $secs[0];
4738                         } else {
4739                                 //track missing section, will replace if found.
4740                                 $rv = $newtext;
4741                         }
4742                 } elseif( $mode == "replace" ) {
4743                         if( $section == 0 ) {
4744                                 $rv = $newtext . "\n\n";
4745                                 $remainder = true;
4746                         } else {
4747                                 $rv = $secs[0];
4748                                 $remainder = false;
4749                         }
4750                 }
4751                 $count = 0;
4752                 $sectionLevel = 0;
4753                 for( $index = 1; $index < count( $secs ); ) {
4754                         $headerLine = $secs[$index++];
4755                         if( $secs[$index] ) {
4756                                 // A wiki header
4757                                 $headerLevel = strlen( $secs[$index++] );
4758                         } else {
4759                                 // An HTML header
4760                                 $index++;
4761                                 $headerLevel = intval( $secs[$index++] );
4762                         }
4763                         $content = $secs[$index++];
4764
4765                         $count++;
4766                         if( $mode == "get" ) {
4767                                 if( $count == $section ) {
4768                                         $rv = $headerLine . $content;
4769                                         $sectionLevel = $headerLevel;
4770                                 } elseif( $count > $section ) {
4771                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4772                                                 $rv .= $headerLine . $content;
4773                                         } else {
4774                                                 // Broke out to a higher-level section
4775                                                 break;
4776                                         }
4777                                 }
4778                         } elseif( $mode == "replace" ) {
4779                                 if( $count < $section ) {
4780                                         $rv .= $headerLine . $content;
4781                                 } elseif( $count == $section ) {
4782                                         $rv .= $newtext . "\n\n";
4783                                         $sectionLevel = $headerLevel;
4784                                 } elseif( $count > $section ) {
4785                                         if( $headerLevel <= $sectionLevel ) {
4786                                                 // Passed the section's sub-parts.
4787                                                 $remainder = true;
4788                                         }
4789                                         if( $remainder ) {
4790                                                 $rv .= $headerLine . $content;
4791                                         }
4792                                 }
4793                         }
4794                 }
4795                 if (is_string($rv))
4796                         # reinsert stripped tags
4797                         $rv = trim( $stripState->unstripBoth( $rv ) );
4798
4799                 return $rv;
4800         }
4801
4802         /**
4803          * This function returns the text of a section, specified by a number ($section).
4804          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4805          * the first section before any such heading (section 0).
4806          *
4807          * If a section contains subsections, these are also returned.
4808          *
4809          * @param $text String: text to look in
4810          * @param $section Integer: section number
4811          * @param $deftext: default to return if section is not found
4812          * @return string text of the requested section
4813          */
4814         public function getSection( $text, $section, $deftext='' ) {
4815                 return $this->extractSections( $text, $section, "get", $deftext );
4816         }
4817
4818         public function replaceSection( $oldtext, $section, $text ) {
4819                 return $this->extractSections( $oldtext, $section, "replace", $text );
4820         }
4821
4822         /**
4823          * Get the timestamp associated with the current revision, adjusted for
4824          * the default server-local timestamp
4825          */
4826         function getRevisionTimestamp() {
4827                 if ( is_null( $this->mRevisionTimestamp ) ) {
4828                         wfProfileIn( __METHOD__ );
4829                         global $wgContLang;
4830                         $dbr = wfGetDB( DB_SLAVE );
4831                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4832                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4833
4834                         // Normalize timestamp to internal MW format for timezone processing.
4835                         // This has the added side-effect of replacing a null value with
4836                         // the current time, which gives us more sensible behavior for
4837                         // previews.
4838                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4839
4840                         // The cryptic '' timezone parameter tells to use the site-default
4841                         // timezone offset instead of the user settings.
4842                         //
4843                         // Since this value will be saved into the parser cache, served
4844                         // to other users, and potentially even used inside links and such,
4845                         // it needs to be consistent for all visitors.
4846                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4847
4848                         wfProfileOut( __METHOD__ );
4849                 }
4850                 return $this->mRevisionTimestamp;
4851         }
4852
4853         /**
4854          * Mutator for $mDefaultSort
4855          *
4856          * @param $sort New value
4857          */
4858         public function setDefaultSort( $sort ) {
4859                 $this->mDefaultSort = $sort;
4860         }
4861
4862         /**
4863          * Accessor for $mDefaultSort
4864          * Will use the title/prefixed title if none is set
4865          *
4866          * @return string
4867          */
4868         public function getDefaultSort() {
4869                 if( $this->mDefaultSort !== false ) {
4870                         return $this->mDefaultSort;
4871                 } else {
4872                         return $this->mTitle->getNamespace() == NS_CATEGORY
4873                                         ? $this->mTitle->getText()
4874                                         : $this->mTitle->getPrefixedText();
4875                 }
4876         }
4877
4878         /**
4879          * Try to guess the section anchor name based on a wikitext fragment
4880          * presumably extracted from a heading, for example "Header" from
4881          * "== Header ==".
4882          */
4883         public function guessSectionNameFromWikiText( $text ) {
4884                 # Strip out wikitext links(they break the anchor)
4885                 $text = $this->stripSectionName( $text );
4886                 $headline = Sanitizer::decodeCharReferences( $text );
4887                 # strip out HTML
4888                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4889                 $headline = trim( $headline );
4890                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4891                 $replacearray = array(
4892                         '%3A' => ':',
4893                         '%' => '.'
4894                 );
4895                 return str_replace(
4896                         array_keys( $replacearray ),
4897                         array_values( $replacearray ),
4898                         $sectionanchor );
4899         }
4900
4901         /**
4902          * Strips a text string of wikitext for use in a section anchor
4903          *
4904          * Accepts a text string and then removes all wikitext from the
4905          * string and leaves only the resultant text (i.e. the result of
4906          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4907          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4908          * to create valid section anchors by mimicing the output of the
4909          * parser when headings are parsed.
4910          *
4911          * @param $text string Text string to be stripped of wikitext
4912          * for use in a Section anchor
4913          * @return Filtered text string
4914          */
4915         public function stripSectionName( $text ) {
4916                 # Strip internal link markup
4917                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4918                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4919
4920                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4921                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4922                 # on how many empty links there are on the page - need to figure that out.
4923                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4924
4925                 # Parse wikitext quotes (italics & bold)
4926                 $text = $this->doQuotes($text);
4927
4928                 # Strip HTML tags
4929                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4930                 return $text;
4931         }
4932
4933         /**
4934          * strip/replaceVariables/unstrip for preprocessor regression testing
4935          */
4936         function srvus( $text ) {
4937                 $text = $this->strip( $text, $this->mStripState );
4938                 $text = Sanitizer::removeHTMLtags( $text );
4939                 $text = $this->replaceVariables( $text );
4940                 $text = preg_replace( '/<!--MWTEMPLATESECTION.*?-->/', '', $text );
4941                 $text = $this->mStripState->unstripBoth( $text );
4942                 return $text;
4943         }
4944 }