includes/Parser.php

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